Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
UNIVERSITÀ DEGLI STUDI DI TRIESTE
XXIX CICLO DEL DOTTORATO DI RICERCA IN CHIMICA
NANOSTRUCTURED MATERIALS FOR ENVIRONMENTAL AND
ENERGY-RELATED APPLICATIONS
Settore scientifico-disciplinare CHIM/03
DOTTORANDO
MATTEO MONAI
COORDINATORE
CHIAR. MO PROF. MAURO STENER
RELATORE
CHIAR.MO PROF. PAOLO FORNASIERO
TUTORI
CHIAR. MO PROF. RAYMOND J. GORTE
DR. TIZIANO MONTINI
ANNO ACCADEMICO 2015 / 2016
- "The most exciting "
Isaac Asimov
i
Table of Contents
0. ABSTRACT ....................................................................................................................................... 1
1. INTRODUCTION ............................................................................................................................... 3
1.1 Emissions Control: Methane Catalytic Oxidation ........................................................................ 4
1.2 Biofuels........................................................................................................................................ 7
1.3 Sustainable Hydrogen Production ............................................................................................... 9
1.4 Aim of the Work and Choice of the Material .............................................................................. 9
2. CHARACTERIZATION TECHNIQUES ................................................................................................. 12
2.1 Physisorption ............................................................................................................................. 12
2.2 Chemisorption ........................................................................................................................... 17
2.3 Powder X-Ray Diffraction .......................................................................................................... 21
2.4 Fourier Transform Infrared Spectroscopy ................................................................................. 22
2.5 Transmission Electron Microscopy Techniques ......................................................................... 23
2.6 X-Ray Absorption Spectroscopy ................................................................................................ 29
2.7 X-Ray Photoemission Spectroscopies ........................................................................................ 35
2.8 Atomic Force Microscopy .......................................................................................................... 38
2.9 Scanning Electron Microscopy .................................................................................................. 39
2.10 Catalytic Activity Measurements ............................................................................................ 42
3. METHANE CATALYTIC OXIDATION ................................................................................................. 48
3.1 Introduction .............................................................................................................................. 48
3.2 Hierarchical Pd@MOx-based Catalysts Synthesis ..................................................................... 50
3.3 Effect of Water .......................................................................................................................... 55
3.4 Phosphorus Poisoning ............................................................................................................... 73
3.5 SO2 Poisoning ............................................................................................................................ 87
ii
4. BIOMASS TO BIOFUELS: HYDRODEOXYGENATION REACTION ..................................................... 112
4.1 Introduction ............................................................................................................................ 112
4.2 H2 Pressure Dependence of HDO Selectivity for Furfural over Pt/C Catalysts ......................... 115
4.3 Mechanisms for High Selectivity in HDO of HMF over Pt-Co Nanocrystals ............................. 124
4.4 Base Metal-Pt Alloys for HDO of HMF..................................................................................... 146
4.5 Ni-Cu Alloys for HDO of HMF .................................................................................................. 158
5. DYE-SENSITIZED PHOTOCATALYTIC H2 PRODUCTION................................................................... 170
5.1 Introduction ............................................................................................................................ 170
5.2 Phenothiazine-based Sensitizers: N-functionalization Effect .................................................. 176
5.3 Carbazole-based Sensitizers .................................................................................................... 186
5.4 Future Perspectives for Sustainable H2 Production ................................................................. 195
6. CONCLUSIONS ............................................................................................................................. 198
7. ACKNOWLEDGMENTS ................................................................................................................. 202
8. BIBLIOGRAPHY ............................................................................................................................ 204
1
0. Abstract
The world is facing an era of global environmental pollution, as a result of the tremendous
population growth and the consequent massive fossil fuel-based energy consumption. A
significant exploitation of renewable energies is needed to guarantee quality of human life and
allow further sustainable growth, but this may take decades to happen. In order to mitigate the
negative effect of human activities on the environment in the short- and mid-term, the
development of more efficient technologies for emissions abatement and for renewable fuels
production is imperative. Heterogeneous catalysis and photocatalysis are two key pillars of a multi-
approach strategy to solve these issues.
During the last century, catalysts were explored by changing the formulation of multi-
component systems in order to find the best performing material for a certain reaction. Since the
late 90's, a new approach to catalytic systems improvement emerged: nano-catalysis. Exploiting
the tools of nanotechnology, tailored nanostructured materials can now be produced, which show
different properties in comparison to their bulky counterparts, often resulting in better catalytic
performances. Furthermore, combining the elements of the periodic table in nano-alloys allows to
expand the possibility of catalyst generation. Consistently with these approaches, the main focus
of this thesis is the synthesis and characterization of well-defined nanostructured and hierarchical
materials for environmental and energy-related applications, such as emissions control, biofuels
synthesis and photocatalytic H2 production. We show that structural control at the nanoscale is a
great instrument for understanding reaction pathways, for studying the nature of catalytic active
sites, and for synthesizing more selective, active and stable catalysts.
Two synthetic strategies were followed to acquire nanostructural control: a self-assembly
method was employed to prepare hierarchical materials starting from functional nanoparticles,
and advanced solvothermal methods were used to prepare monodisperse nanocrystals having
controlled size and composition. State-of-the-art hierarchical Pd-based catalysts embedded by
metal oxide promoters were tested for methane catalytic oxidation in the presence of poisoning
compounds typically found in real applications. Detailed surface studies allowed to propose
deactivation mechanisms and strategies to improve catalysts resistance to deactivation. Well-
controlled nanostructured Pt-based alloys and Ni-Cu alloys showed improved activity, stability and
selectivity for hydrodeoxygenation reactions of biomass-derived feedstocks to produce biofuels.
The control of nanostructure was pivotal to understand the reason for such enhanced
2
performances. Finally, dye-sensitized photocatalysts were investigated in H2 photocatalytic
production under visible light, and state-of-the-art stability and activities were demonstrated.
All these findings greatly contributed to the development of catalytic materials for energy-
related applications.
3
1. Introduction
Since the Great Acceleration of the 20th century, human activity has modified the
environment on so many levels and to such an extent that we may have entered a new geological
epoch - Anthropocene - as proposed by a growing number of researchers (1). As sensationalist as it
may sound, the evidences of anthropogenic impact on the environment are striking, not only on a
geological scale: climate forcing, rising sea levels, increasing CO2 and CH4 concentration in the
atmosphere, erosion of the ozone layer and oceans acidification are just some examples of the fast
transformations taking place since the last century. In response to this, a growing awareness about
the fragility of the biosphere culminated in environmentalism as a political movement, and in the
development of environmental laws and agreements meant to address pollution on an
international scale (2). In this context, environmental science came alive as an active field of
scientific investigation, focusing on pollution abatement, sustainable growth, technological impact
on the environment and many other subjects.
The main thrust for the Great Acceleration came from the great progress in industrial
catalysis of that time. The Haber-Bosch process for ammonia synthesis allowed the production of
economic fertilizers and in turn boosted the growth in population. Later on, the industrial catalytic
production of plastics and rubbers introduced completely new materials to the public
consumption. The invention of the internal combustion engine and the widespread of cars pressed
for petroleum industry growth and new catalytic systems to produce fuels and control noxious
emissions. The dawn and rise of the electronics industry called for new materials and for
miniaturization of integrated circuits, boosting research in optics, electronic properties of materials
and nanostructured materials manufacturing. Aviation and space exploration stimulated research
on innovative materials and on the control of their micro and nanostructure. All these innovations
are tightly linked to scientific advances, and have lead to tremendous developments in
microscopy, spectroscopy, crystallography and many other techniques.
Now, catalysis and material science are dealing with another main issue, which is the
energy problem. Non-renewable fossil fuels such as coal, petroleum and natural gas make up more
than 80 % of the primary energy consumption worldwide (3). The use of fossil fuels raises many
environmental concerns, since their reserves are distributed unevenly on the planet, their
extraction has severe impact on the soil, and because they produce CO2 when burned, which in
turn is the main greenhouse gas contributing to global warming. Combustion of fossil fuels also
4
produces other air pollutants, such as nitrogen oxides (NOx), SO2, volatile organic compounds
(VOCs), and heavy metals.
Despite of these drawbacks, fossil fuels are still the most economic and widespread energy
source, since renewable sources (such as hydroelectric, solar, wind, and geothermal) currently
require more expensive production and processing technologies. Given the inertia of large energy
systems, a significant shift to renewable sources may take decades. In order to mitigate the short
and mid-term environmental impact of fossil fuels, two parallel strategies can be envisaged: the
development of more efficient emissions control/storage systems and a shift towards more
sustainable biofuels and energy carriers. These tasks are very complex, and no single strategy can
be applied to answer the problem. However, catalysis, and heterogeneous catalysis, are the
common keys to achieve such a target, thanks to recent advances in nanotechnology.
Nanotechnology is a field of research that encompasses science and engineering, and
refers to the manipulation of matter on an atomic and molecular scale. Nanotechnologic materials
are already finding applications in electronics, biology, materials science, medicinal chemistry and
catalysis, and they can be used to exploit energy production processes in a more sustainable way
than the present ones, also in terms of pollutant emissions. For what heterogeneous catalysis is
concerned, many nanotechnologic materials have already been proved to be more active and
selective than their classical counterparts in the desired catalytic process, such as hydrogen
production (4, 5), steam reforming (4), catalytic oxidation (6 8), etc.
Accordingly, the present work is focused on the synthesis and characterization of well-
defined nanostructured and hierarchical materials for environmental and energy-related
applications, such as emissions control, biofuels synthesis and photocatalytic H2 production. We
show that structural control at the nanoscale is a great instrument for understanding reaction
pathways, for studying the nature of catalytic active sites, and for synthesizing more selective,
active and stable catalysts. The aim of this introductory chapter is to underline the objectives of the
present study and to provide the reader with a general survey of the matter of interest of this
thesis.
1.1 Emissions Control: Methane Catalytic Oxidation
Emissions control is not a new concept: since the original Clean Air Act1 of the '70s, catalytic
technologies for noxious emissions abatement substantially improved the quality of life, especially
in urban areas (9), where the emission levels of hydrocarbons (HCs), carbon monoxide (CO) and
nitrogen oxides (NOx) from automobiles have decreased by more than 90% (in 2004) (10).
1 The Clean Air Act excludes methane from the emission standards, since it does not significantly
participate in photochemical reactions.
5
Nowadays, clean air is considered to be a fundamental requirement for human health and
wellbeing. However, the challenge remains to reduce the emission levels of pollutants such as
volatile organic compounds (VOCs), particulate matter (PM), ammonia, nitrogen oxides and ozone
(O3) even further. These compounds arise (directly or indirectly) from natural sources and from
many human activities, comprising agricultural activity, combustion processes, transportation,
solvent usage, industrial processes, etc. (11). Transportation is one of the main sources of air
pollution in Europe, particularly in cities and urban areas such as towns, airports and sea ports (12).
According to the revised Gothenburg Protocol (4 May 2012)2, between 2005 and 2020 the EU
member states must jointly cut their emissions of volatile organic compounds by 28%, together
with sulfur dioxide by 59%, nitrogen oxides by 42%, ammonia by 6%, and particles by 22%.
In this work, we focus on the catalytic abatement of methane (CH4), one of the most
climate-impacting VOC. VOCs are organic species that evaporate steadily at room temperature and
participate in atmospheric photochemical reactions, excluding carbon monoxide, carbon dioxide,
carbonic acid, metallic carbides or carbonates, and ammonium carbonate3. Because of the very
large number of individual air pollutants that come within the definition of VOCs, their importance
as a class of ambient air pollutants has only recently become recognized (13).
After carbon dioxide, methane is the most important contributor to radiative climate
forcing, especially if including its indirect effect due to chemical reaction with hydroxyl radicals
(·OH) in the atmosphere (14). Methane concentrations in the atmosphere have raised exponentially
in the last century, going from a preindustrial concentration near 700 ppb to a present-day value of
1745 ppb (13, 14). In line with this, satellite retrievals and surface observations have shown that U.S.
methane emissions increased by more than 30% over the 2002 2014 period (15). However, this
increase cannot be readily attributed to any specific source type, and the Environmental Protection
Agency (EPA) indicate no significant trend in U.S. anthropogenic methane emissions from 2002 to
present. This may be due to the fact that methane is a well-mixed pollutant globally, so that
ambient concentrations may not strictly be related to regional emissions. New worldwide
regulations are therefore needed, revising the proposals of the Kyoto Protocol and the United
Nations Framework Convention on Climate Change (UNFCCC).
Many technologies for VOCs abatement have been developed for a wide range of
applications, such as solids adsorbents and other capture devices, condensers, membranes,
biodegradation methods and thermal or catalytic oxidation (16). However, each technology can be
2 Protocol was finalized at a meeting of the parties to the Convention on Long-range Transboundary
Air Pollution (CLRTAP) in Geneva. http://www.airclim.org/acidnews/new-gothenburg-protocol-adopted 3As of 03/29/2013, Electronic Code of Federal Regulations, 40 CFR 51.100(s) - Definition - Volatile organic compounds (VOC). http://www.ecfr.gov/cgi-bin/text idx?c=ecfr&rgn=div8&view=text&node=40:2.0.1.1.2.3.8.1&idno=40
6
applied for a specific range of organic compounds, concentrations, and emission sources (17, 18).
For example, low-temperature condensation is energy intensive and limited to treatment of
evaporative solvents, while biochemical methods are selective and concentration sensitive. One of
the most versatile and economical VOC removal technologies is catalytic oxidation (19), because it
can operate with dilute VOCs effluent streams (<1% VOCs) and at much lower temperatures and
higher selectivity than conventional thermal incineration, with consequent limited emissions of
NOx. Some limitations of this abatement technique include the disposal of the spent catalyst
materials (if not recyclable), catalyst poisoning by non-VOC materials, aging and leaching of the
catalyst (18). Also, to date no commercial technology exists able to control methane emission at
medium-low temperature (350-400°C). Therefore, the development of new effective methods for
VOCs abatement, and methane in particular, is mandatory.
Presently, between 50 and 90% of the total HCs emissions from modern cars equipped with
three-way catalysts (TWC) are released during the cold start, i.e. when the engine is still cold and
the catalyst temperature is low (10). Indeed, the most critical factor for efficient conversion is the
temperature of the catalyst. After start-up, the hot engine exhausts heat up the catalyst to a
temperature high enough to initiate the catalytic reactions, called the light-off temperature, which
sets an approximate limit between the kinetically and mass transport controlled temperature
regimes for the catalytic reactions. During low speed driving condition CH4 emissions are 5 times
as much as the emissions at high speed (20). Even in the case of natural gas vehicles (NGVs), that
offer significant environmental advantages over gasoline and diesel (21), the problem of unburned
methane emissions represents a serious ecological hazard, especially during the sold start (22).
Indeed, methane is the principal hydrocarbon species emitted by NGV, reflecting the fuel
CH4/NMHC (non-methane hydrocarbons) ratio, typically 93% methane, 5% NMHC(23).
A number of approaches have been proposed to improve the emission control during cold-
start, such as increasing catalysts activity and developing technical solutions that more indirectly
improve the efficiency of the abatement system. The challenge remains to reach the working
temperature of the catalyst as fast as possible, and nanotechnologic materials may have the
answer to this problem, as demonstrated by our group of research with the development of a
nanostructured catalyst exceptionally active in the catalytic oxidation of methane at low
temperature (8), based on Pd@CeO2 units supported on a high-surface area Al2O3-based support.
In Section 3 of the present work, we study the effect of poisoning compounds typically
found in real applications on the performance of Pd@MOx (MOx= CeO2, ZrO2 or Ce-Zr mixed oxides)
nanostructured catalysts in the methane catalytic oxidation reaction. Detailed surface studies
7
allowed to propose deactivation mechanisms in the presence of H2O, phosphates or SO2, and to
develop strategies aimed to improve catalysts resistance to deactivation.
1.2 Biofuels
World transportation fuel consumption is expected to double by 2035 under a business-as-
usual scenario, with forecasted CO2 emissions reaching 12 gigatonnes per year (Gt yr-1) (24). In
order to reduce the environmental impact of such a staggering growth, the energy efficiency of
conventional fuels should be increased, and alternative fuels (e.g. biofuels, electricity) should be
extensively used.
Currently, 95% of transportation fuels are being produced from crude oil. One of the most
mature energy alternatives to fossil fuels are biofuels, i.e. fuels obtained by upgrading biomass
feedstocks. Biofuels production processes are not without problems, but they are currently in
practice around the world. For instance, in 2013 the US consumed more than 14.5 billion gallons of
biofuels, reducing the need for gasoline and diesel fuels made from crude oil (25). Recent reports
claim that cellulosic biofuels can reduce greenhouse gas emissions by more than 60% compared to
petroleum-derived gasoline4 (26). This means that the same amount of energy can be produced
with significantly less climate-disrupting pollution.
Biofuels are generally classified as first, second or third generation biofuels, depending on
the feedstock and the processing technology used (27). Most biofuels produced worldwide are
first-generation, that is, derived from food crops by fermentation to bioethanol or by abstraction of
oils for use in biodiesel (26). However, first-generation biofuels present serious limitations, such as
the competition with food sources exploitation, so research and development is focused on
second generation biofuels, i.e. derived from non-food, lignocellulosic sources, such as agricultural
and forest wastes, energy crops, and even algae (third generation biofuels).
The main issues impeding widespread low-cost production of biofuels from lignocellulosic
biomass are its recalcitrant nature and the diversity of chemico-physical properties of its
components. These properties vary dramatically not only among feedstocks but even for a certain
kind of feedstock, to the point that the US Department of Energy (DOE) has established a Bioenergy
Feedstock Library, a database for physical, chemical and conversion performance characteristics of
more than 50,000 biomass feedstocks5. This complex scenario is further worsen by the expensive
pretreatments and upgrading processes required for both biochemical and thermal conversion of
biomass to biofuels (24). 4 To assess the life-cycle impacts of biofuels on greenhouse gas emissions, biofuels are credited with
all the carbon dioxide captured and stored as the biomass grows, as well as any emissions created during harvest, conversion, distribution, and use.
5 https://bioenergylibrary.inl.gov/Home/Home.aspx
8
In order to achieve economically and environmentally sustainable production of biofuels,
advances in several fronts of plant science, biochemical and chemical engineering, catalysis and
analytical chemistry are needed. Agronomic science advances could yield tailored energy crops
characterized by enhanced plant productivity, efficient nutrient utilization, reduced water demand
and higher resistance to pests and disease, while providing favorable life-cycle benefits. From a
biochemical perspective, genome editing techniques can be used to manipulate plants
components (such as lignin) on a molecular level, in order to enhance our ability to convert
biomass to fuels, by making biomass easier to break down. For instance, research at the Bioenergy
Science Center6 proved that genetic modification of switchgrass can produce phenotypically
normal plants having reduced thermal- c, and microbial recalcitrance,
showing the potential to lower biomass processing costs (28).
Progress in chemical engineering is also essential for improving biofuels processing
technologies. There is a need for lower-cost, better performing catalysts for hydrodeoxygenation
(HDO) reactions and hydro-treatment processes along with the ability to produce the hydrogen
needed for upgrading in sustainable ways. Concurrently, engineering shall deliver advances in
separation technologies, reactor design, sensors, process control and waste minimization and
treatment (24). For biofuels to be competitive with fossil fuels, their production facilities should
progress from a single-product the bio-refinery model, in which starting resources are fractionated
and all molecular components are used to maximize value generation.
Section 4 of this work is focused on HDO, an important step of thermo-chemical biofuels
production. HDO is a hydrogenolysis process for removing oxygen from oxygen-containing
compounds, especially relevant to the upgrading of bio-oils and lignocellulosic-derived feedstocks
(29). Despite the extensive literature on the subject, many aspects remain unclear, such as the
influence of reaction conditions, reactant molecule and catalyst composition and nanostructure on
the catalytic performance. Herein, we first investigate and compare the mechanism of reaction of
two important model compounds (furfural and 5-hydroxymethylfurfural ─ HMF), usually studied
separately and with different approaches, to show that they share a very similar reaction pathway
(Section 4.2) (30). We then report the effect of alloying Pt with non-noble metals (Co, Cu, Ni, Zn),
and of alloying two base metals (Ni-Cu) on the catalytic performances in HDO of HMF, showing that
for certain optimal composition the alloyed materials achieve much improved selectivity and
stability with respect to single metal catalysts (Section 4.3-4.5). The nanostructural control over the
studied catalysts, coupled with advanced spectroscopic and imaging techniques, allowed for a
rationalization of the observed performances in terms of catalysts structure and composition.
6 Funded by the US DOE, http://bioenergycenter.org/besc/
9
1.3 Sustainable Hydrogen Production
H2 is a pivotal chemical in the industry and energy sectors, with applications in petroleum
refining, power generation in turbines and fuel cells, methanol and ammonia production, welding
and heat-treatments, semiconductor manufacturing and hydrogenation of unsaturated fatty acids
in vegetable oil, to name a few. More than 50 million metric tons of H2 are produced annually
worldwide7, mainly by steam reforming of natural gas. Sustainable H2 production from low-carbon
or renewable energies remains challenging and will require large-scale changes to our energy
systems (24).
The existing methods for the production of renewable hydrogen can be divided into two
major categories: direct solar-to-H2 conversion and electrolytic conversion methods (31). In
electrolytic routes, the renewable energy source is converted to electricity, and H2 is then produced
via electrolysis of water. In solar-driven water-splitting systems, photons instead of electricity are
used to convert water into hydrogen and oxygen with the use of photocatalysts. Further advances
in materials chemistry, catalyst development and system engineering are required for these
systems to be competitive with existing technologies. For instance, there is a need for low-cost
materials exhibiting high energy-conversion efficiencies, and for a collection and distribution
system for hydrogen, comparable to landfill gas collection and piping. Although a scalable system
has yet to be shown, significant progress has been made on each of the necessary components of a
demonstration system (24).
Section 5 of this work is focused on direct solar-to-H2 conversion over dye-sensitized
photocatalysts. Sensitization of TiO2 with visible light-absorbing moieties such as colored dyes is a
widely investigated strategy to enhance the photocatalytic efficiency of TiO2 by extending its
otherwise UV-limited light absorption to the visible (Vis) range (32). In this thesis, we report
advances in the synthesis of stable and efficient dye-sensitized Pt/TiO2 through the rational
molecular design of sensitizers. State-of-the-art performances are achieved using the synthesized
dyes coupled with benchmark materials, and strategies to further improve the process
sustainability are outlined.
1.4 Aim of the Work and Choice of the Material
This work focuses on some major aspects of the energy problem: emissions control,
biofuels production and H2 renewable photocatalytic production. Well-defined nanostructured
materials are developed, characterized and tested for catalytic reactions, with the aim of
rationalizing catalytic performances in terms of structure, composition and surface chemistry.
7 https://www.hydrogen.energy.gov/
10
Three main classes of materials are investigated: hierarchical materials based on Pd@MOx
units for methane catalytic oxidation, Pt-based nano-alloys (of Co, Ni, Zn, Cu) and Ni-Cu nano-alloys
for hydrodeoxygenation reactions, and dyes-sensitized Pt/TiO2 for H2 photocatalytic production
under visible light.
Pd@MOx-based (M=Ce, Zr, or Ce-Zr) catalysts supported on high surface area Si-Al2O3 are
selected for methane oxidation studies because they show state-of-the-art activity at low
temperature under ideal conditions and high thermal stability, as previously reported by our
research group (8). The focus of this section is the study of Pd@MOx-based catalysts performances
under conditions approaching the ones of real applications, such as industrial and automotive
exhausts. H2O, phosphates and SO2 are chosen to test the catalysts resistance because they are the
most powerful poisoning and deactivating agents of Pd-based catalysts (33).
Model catalysts based on Pd@MOx units supported on conductive materials are prepared
for advanced photoelectron spectroscopy (PES) investigations, in order to gain insights into the
transformation occurring at the catalysts surface during deactivation and to develop strategies for
the synthesis of more resistant second-generation catalysts. In the P-poisoning study, graphite is
used as a support because of its high conductivity and the possibility to deliberately introduce P in
the system during the catalyst preparation (34 36), rather than introducing it from the gas feed
during aging treatments (37). In this way, H2O is not introduced in the reaction mixture by
decomposition of H3PO4 to P2O5 and the effect of water addition can be studied separately. For the
SO2 poisoning study, indium tin oxide (ITO) is used as a conductive support because of its
conductivity and high thermal resistance under oxidative conditions. Atomic Layer Deposition
(ALD) is then used to deposit thin Al2O3 films on the ITO surface, in order to produce a support
having a surface chemistry similar to high-surface area Al2O3, while retaining the conductivity
required for precise PES studies.
Pt-based and Ni-Cu nano-alloys are selected for HDO studies because of the considerable
evidence that bimetallic catalysts can be more selective for HDO reactions (38). The aim of this
work is to understand the reason for such enhanced selectivity and to investigate the nature of the
catalytic active sites for HDO reactions. Solvothermal methods are used to synthesize the
nanocrystalline alloys in this work in order to achieve nanocrystals having well-controlled and
uniform shape, size and composition (39, 40). Uniformity is critical to study the factors influencing
selectivity, especially for alloys in which both the metals are able to catalyze the HDO reaction, such
as Pt-Co and Pt-Ni. Conventional methods, such as impregnation, are not able to produce such
uniformity, which is the reason we prepare the catalysts in this study by synthesizing uniform NCs
in solution.
11
The HDO catalytic studies reported in this thesis are carried out in a high-pressure, three
phase tubular-flow reactor, because such setup allows to easily change experimental parameters
(e.g. space velocity) and investigate their effect on the performance of catalysts. Tubular-flow
reactors are preferred to batch reactors, for which the contact between the solid catalyst, the
liquid-phase reactant, and the gas-phase H2 can be ineffective and activity can be limited by H2
diffusion (41). Also, in batch reactors, reactions can occur during the heating and cooling periods,
further complicating the interpretation of results. In the first part of the HDO study, the importance
of reaction conditions is underlined in the case of furfural, which was studied for the first time
under high-pressure liquid flow conditions and compared to the widely reported low-pressure, gas
phase setup, in order to get new insights in the HDO reaction mechanism.
For dye-sensitized H2 photocatalytic production studies, Pt/TiO2 nanomaterials are used as
a benchmark because of their well-known photocatalytic properties (42). The performance of the
dyes as sensitizers is evaluated by studying the H2 photocatalytic production over dye-stained
Pt/TiO2 under visible light irradiation (λ > 420 nm). The aim of this study is to enhance
photocatalytic performances of dyes/Pt/TiO2 systems through the rational molecular design of
donor-acceptor dyes. In particular, the effect of peripheral functionalization (aimed to increase
surface wettability) and heteroatom substitution in the D core and π-spacers are discussed for a
series of dyes previously reported for dye-sensitized solar cells applications (43, 44).
The reaction conditions are also carefully optimized in terms of dye loading and amount of
catalyst in the reaction suspension, following recent guidelines for photocatalytic studies (45).
Triethanolamine (TEOA) is used as sacrificial agent in most of the reactions because it is widely
reported in the literature (46), even if it's not a very sustainable choice. As a future perspective, the
H2 sustainable photocatalytic production from EtOH/water mixtures will be finally presented.
12
2. Characterization Techniques
In the following section an overview of the experimental techniques used in the present
study is reported. Different techniques were employed to characterize the prepared catalysts and
photocatalysts in terms of textural, structural and morphological properties. Their catalytic activity
was evaluated for various reactions, relevant in the general context of sustainable energy
production and environmental protection, such as catalytic oxidation of methane, biomass
upgrading to biofuels and photocatalytic hydrogen production. A brief summary of each technique
is given, highlighting the main elements that will help evaluating the results obtained in this thesis.
2.1 Physisorption
The activity of heterogeneous catalysts is strictly related to the morphology and the
extension of their surface area. Indeed, gas reactions catalyzed by solid materials take place on
those active sites of the catalyst that are in contact with the reactants phase, i.e. that are located on
the catalyst interior or exterior surfaces accessible to the reactants. If the pores are wide enough to
permit the diffusion of reactants and products to and from the active sites, the activity is directly
proportional to the number of active sites. This principle is valid for all heterogeneous catalytic
reactions, allowing to study the reactions under kinetic control (47). Otherwise, if pores are too
narrow, the reaction is limited by the mass transport of reactants from the gas phase to the active
site, and it is said to be under diffusional control. In this case the activity could be independent of
the surface area or proportional to its square root, depending on the mode of diffusion. It is
deduced that the catalytic activity of a solid material is not only influenced by its surface area, but
also by its pore structure (texture and dimensions of the pores). For example, a particular pore
structure may induce a shape selectivity to a reaction, limiting the diffusion of one particular
reactant or product (47).
The activity of heterogeneous photocatalysts is also strongly dependent on surface area,
since photocatalytic reactions occur on the material surface. Large surface areas provide more
active sites to react with absorbed water and hydroxyl species to form hydroxyl radicals and anchor
organic molecules for photodegradation (48). However, performances is also limited by the surface
area available for capture of solar light and by the distribution and density of surface localized trap
states at grain boundaries, which cause recombination of electrons and holes and depend on the
material crystallinity and morphology (49). To obtain highly active TiO2 photocatalyst, therefore, it
is important to simultaneously consider these properties.
13
Even in the same material, pores can vary in shape and dimension in a wide range. Pores
can be divided into three groups on the basis of their dimensions:
micropores are pores having diameters smaller than 2 nm;
mesopores have diameters in the range 2 - 50 nm and are typical of non crystalline
materials;
macropores have diameters larger than 50 nm.
In the particular case of core-shell modular units supported on alumina, the pore size
distribution of the support becomes very important because pores narrower than 20-30 nm are too
small to be accessible to the core-shells, and so the associated fraction of surface area is not
available for their deposition (8). The ability to measure the surface area of a catalyst and the
dimension and distribution of its pores is therefore essential to any catalytic study.
2.1.1 Estimating Surface Area
Among the well-established methods that are used to determine the surface area of a
porous solid, the most commonly employed is the volumetric method. This method is widely used
for determining the surface area and pore size distribution of a variety of different solid materials,
such as ceramic, industrial adsorbents and catalysts (50, 51). The volumetric method consists in
measuring the adsorption of an inert gas on the solid at a given constant temperature as a function
of the partial pressure of the adsorbent. Notably, a fundamental requisite to apply this technique is
that the interaction between the adsorbent molecules and the material surface has to be weak.
Therefore, only a physical interaction must take place. This process allows to obtain a physisorption
isotherm. On the basis of the isotherm shape, it is possible to determine the surface area and pore
distribution of the sample, according to empirical equations and adsorption models (47).
Since physisorbed molecules are not restricted to specific sites but are free to completely
cover the surface of the solid, the method allows to estimate the total surface area of the sample.
Furthermore, since the interactions that lead to the adsorption are reversible and weak, the process
doesn't modify the surface of the sample, so is not invasive. The process is also reversible, so both
adsorption and desorption processes can be studied. Moreover, because many molecular layers of
adsorbate can be formed, the pore volume may be measured if the amount of adsorbate needed
to completely fill the pores can be extrapolated. According to the International Union of Pure and
Applied Chemistry (IUPAC) recommendations (50), the majority of physisorption isotherms can be
classified into six types, that are displayed in summarized in Figure 2.1.
14
Figure 2.1 Physisorption isotherm types and hysteresis classification according to the
IUPAC recommendations (Adapted from (50))
The standard physisorption isotherms can be described as follow:
Type I isotherms are given by microporous solids having relatively small external surfaces,
such as activated carbons and zeolitic molecular sieves.
Type II isotherms are typical of non-porous or macroporous solids, on the surface of which
unrestricted monolayer-multilayer adsorption takes place.
Type III isotherms are not common. They are convex to the relative pressure axis over their
entire range and, as we shall discuss later, are not useful to extrapolate the surface area of
the solid. The adsorbate species may be replaced by another one in order to get another
type of isotherm.
Type IV isotherms are given by many industrial adsorbents. They present a hysteresis loop,
which is associated with capillary condensation taking place in mesopores and can vary to
a great extent depending on pores geometry. The hysteresis loop is also very important for
determining the pore distribution of the material. According to the IUPAC
recommendations, these are grouped in four types (H1-H4 in Figure 2.1), basing on their
shape. H1 type is observed for compact agglomerates of spherical particles with rather
uniform dimensions and disposition, while H4 type is observed for adsorbents made of
agglomerates of bi-dimensional particles. H2 and H3 hysteresis types are observed for
15
intermediate situations. Hysteresis are often not easily classified, as most of the materials
show heterogeneous distributions in shapes and dimensions of pores.
Type V isotherms are uncommon and related to type III isotherms: the adsorbent-adsorbate
interaction is weak.
Type VI isotherms represent stepwise multilayer adsorption on a uniform non-porous
surface.
In order to estimate the total surface area of a material using the volumetric method, the
completion of an adsorbed monolayer has to be detected from the isotherm shape by some
means. This is possible only for isotherm types I, II and IV. For isotherm type I, the adsorption is
usually described adequately by the Langmuir equation.
(2.1)
where p is the pressure of gas, b is a constant derived from kinetic principles that depends
on temperature and adsorption heat, V is the volume of adsorbed gas and Vm is the volume of
adsorbed gas forming a theoretical monolayer on the surface of the material. In order to determine
the monolayer volume from empirical data, the Langmuir equation can be rearranged into:
(2.2)
Therefore, a plot of p/V vs. p will give a straight line, the slope of which is 1/Vm, the inverse
of the monolayer volume.
For isotherm types II and IV instead, the Langmuir equation doesn't hold through because
multilayer coverage takes place. The monolayer coverage may be extrapolated roughly by referring
to the ordinate value of the inflection of the isotherms, known as the point B (see Figure 2.1). The
point B method may be used only if the beginning of the almost linear section of the isotherm is
well defined, that is, if a sharp change of curvature is noted. Otherwise, a more sophisticated
analysis of the isotherm is required and the surface area can be calculated by extrapolation of the
monolayer volume applying the Brunauer-Emmett-Teller (BET) theory. The BET model starts from
the assumption that adsorption is a reversible process consisting in the formation of a series of
layers, where the most external one is formed by adsorbate molecule directly in equilibrium with
the vapor phase. On the basis of these principles, the BET equation is derived:
16
(2.3)
where p is the gas pressure, p0 the saturated vapor pressure of the liquid at the operating
temperature, V is the volume of adsorbed gas, Vm is the volume of monolayer coverage and C is the
BET constant (that depends on temperature and interactions between adsorbent and adsorbed
species). Plotting p/V(p0-p) vs. p/p0 usually gives a straight line in the range of 0.05 < p/p0 < 0.35,
and from the slope and intercept the monolayer coverage volume is determined.
Once the monolayer coverage volume Vm is known, the available surface area is calculated
from the equation:
(2.4)
where Vmol is the molar volume of the adsorbate (at the same temperature and pressure of
Vm), NA is the Avogadro's number and am is the area of an adsorbed molecule (am = 0.162 nm2 for N2
at the liquid nitrogen temperature).
2.1.2 Estimating Pore Volume and Diameter
Many methods have been developed to estimate pores distribution of solid materials. In
this study, the gas adsorption method will be employed. This method is based on the
physisorption process, just like the volumetric method used to assess the surface area of solid
materials (47). In this case, however, one wants to observe just one particular phenomenon of the
physisorption process, that is the capillary condensation of the gas used as adsorbate. The capillary
condensation takes place in narrow pores at a lower pressure than the saturation pressure of the
adsorbate. The capillary condensation in cylindrical pores is described by the Kelvin equation:
(2.5)
which is obtained by equating the work spent in enlarging a spherical drop of liquid to that
done in adding molecules to the interior of the drop. In this case, p0 is the vapor pressure at the
f the liquid , r is the radius
of the pore, R is the gas constant and T is the absolute temperature of operation. Basing on this
17
equation, a pore size distribution curve can be constructed, plotting the volume of adsorbed gas
vs. the diameter of the pores calculated for each p/p0.
From this simple equation, it is also possible to understand that the formation of the
hysteresis loop in the physisorption isotherm is related to particular shapes of the pores. In fact, in
the case of cylindrical pores, the capillary condensation takes place at the same partial pressure
during both the adsorption and desorption processes. The hysteresis loop is originated by the fact
that, during desorption, evaporation of the gas takes place at a lower p/p0 with respect to the
condensation during adsorption. Therefore, evaporation takes place on pores with a diameter
smaller than that of pores in which capillary condensation occurs. This situation can be rationalized
only assuming the formation of neck-bottle pores, in which the apertures have diameters smaller
than the cavities (47).
The Kelvin equation is a powerful method for investigating the pore distribution of a
sample, but is insufficient in order to interpret correctly the experimental results obtained via the
gas adsorption method. A model of the porous structure of the material is needed. Many different
pore models have been developed, some qualitative and some much more complicated based on
mathematical simulations. One of the differences among them arises from the pores dimension
range in which they can be applied. In the present work particular attention is given to mesopores,
while micropores are not taken into account because of their scarce thermal stability (they collapse
at typical operation temperatures considered in this study) and because they are too narrow in
diameter to be accessible by core-shell nanostructures (8). The method commonly used to describe
mesopores (and small macropores) distribution is that developed by Barret, Joyner and Halenda
(BJH method) (52). Analyzing the physisorption isotherm in the 0.40 < p/p0 < 0.98 range (region
comprised between the formation of the monolayer and the saturation of the system, equivalent
to a complete filling of pores), it is possible to obtain the mesopore volume and pore distribution.
In the present work, N2 physisorption experiments were carried out on a Micromeritics
ASAP 2020C. The samples were first degassed in vacuum at 350 °C overnight prior to N2 adsorption
at liquid nitrogen temperature.
2.2 Chemisorption
Chemisorption is based on a specific interaction between a probe molecule and the metal
surface that constitute the active phase of a catalyst (47). The probe molecule is able to chemically
react with the metal, producing a single layer of chemisorbed molecules. The measure of the
volume of gas consumed for the formation of the monolayer allows the calculation of the area of
the active phase and consequently its dispersion. A typical chemisorption isotherm is shown in
Figure 2.2.
18
Figure 2.2 Typical chemisorption isotherm.
The number of surface metal atoms N(s)M and the active metal surface SM can be obtained
from the following equations:
(2.6)
(2.7)
where V is the volume of adsorbed gas, Vm is the gas molar volume, NA is the Avogadro's
number, aM is the cross sectional area of a single metal atom and n is the chemisorption reaction
stoichiometry, that is the number of metallic atoms that are needed to bind a single molecule of
adsorbate.
The most common gases used for chemisorption studies are hydrogen and carbon
monoxide. Notably, many other gases can be used as probe for chemisorption. The best choice of
the gas depends on the nature of the metal and of the support included in the formulation of the
catalyst under analysis.
Generally, the chemisorption stoichiometry with hydrogen is assumed to be 2, because H2
can be activated by the metal surface and subsequently dissociated, each hydrogen atom forming
a bond with one metal atom. On the other hand, the stoichiometry of adsorption CO on metals
(such as Pd and Pt) is assumed to be 1. These assumptions are not always true: actually, for very
19
small nanoparticles the stoichiometric coefficient of CO may vary, because of the formation of
geminal di-carbonylic species on edge and vertex atoms (53). Similarly, in the case of H2 the
stoichiometric coefficient may vary for extremely dispersed particles (lower than 1nm), but also the
formation of metals hydrides and spillover effects (especially in reducible oxides) can alter the
measurement (54). These phenomena must be taken into account during planning of the
measurements and evaluating the experimental results.
In the case of non-dissociative adsorption (as is the case of CO) the Langmuir isotherm
equation can be applied, assuming a constant chemisorption energy:
(2.8)
where nads is the quantity of gas adsorbed at pressure p, nmads is the quantity of gas needed
for the formation of the monolayer and b is a constant. The saturation limit should correspond to a
horizontal section of the isotherm in the high pressure region. However, as shown in Figure 2.2 this
is not usual because, in most of the experiments, this region shows a straight positive trend due to
the formation of subsequent layers due to the physical adsorption of the probe gas. The monolayer
volume may be calculated by extrapolating the linear part of the isotherm to zero pressure, but a
more elegant method exists. This is based on the subtraction of the physisorption contribute from
the chemisorption isotherm: the total isotherm is measured, then the system is evacuated at the
temperature of analysis for a short time, leading to desorption of the physisorbed gas, and finally
another isotherm is measured. The difference between the two gives the irreversible adsorption,
which is due solely to the contribution of chemisorbed species. The monolayer adsorption is then
calculated extrapolating to zero the linear section of the subtracted isotherm.
From the chemisorbed volume it is possible to obtain the metal dispersion DM and the
average metal particle diameter dM. To obtain these values, the geometry of the metal particles has
to be assumed. Assuming a geometrical shape of the particles:
(2.9)
(2.10)
20
where N(s)M is the number of surface metal atoms, N(tot)M is the total number of metal atoms,
6 is the geometrical factor for spherical particles and aM is the area of a metal atom. VM is the
volume of a bulk metal atom. It can be calculated from the Equation 2.11:
(2.11)
density of the metal.
In the present thesis, chemisorption experiments were performed on a Micromeritics ASAP
2020C. The samples (~ 150mg), placed in an U-shaped quartz reactor, were subjected to
preliminary thermo-chemical treatments in order to clean and reduce the sample and finally to
fully eliminate the hydrogen gas from the surface of the catalyst. Usually, samples were cleaned by
a treatment under flowing O2(5%)/Ar at 500°C for 1h, reduced under H2(5%)/Ar at 80°C for 1h and
finally evacuated at 350°C for 4h.
For any Pd/CeO2 system there are some particular elements that should be underlined.
First, the support has the ability to spill-over the hydrogen. To obtain the chemisorption
contribution of the active phase alone it is thus necessary to halt the spill-over phenomenon. It has
been observed that it is possible to do so by measuring the chemisorption isotherm at low
temperature (solid/liquid acetone cooling bath, -90°C) (54). At this temperature, the kinetic of
hydrogen diffusion on the oxide surface is very slow, while the adsorption kinetic is almost the
same (since the activation energy is almost zero). However, operating at low temperature results in
a much greater physisorption contribution. The chemisorbed hydrogen can be obtained via
extrapolation at p=0 of the linear part of the isotherm measured at 180 K. The isotherms measured
at room temperature allow instead to compare the mobility of hydrogen on the oxide surface.
Another aspect to be considered is that the palladium is known for its ability in absorb hydrogen,
leading to the formation of hydrides. Therefore, chemisorption studies of the Pd/CeO2 catalysts are
usually performed at low temperature (about -90 °C) and low H2 pressures (2 - 20 Torr) (54).
To avoid these complications, in this study CO chemisorption experiments were conducted.
Whereas with some metals, such as nickel, the adsorbed CO molecule may thermally dissociate on
the surface, with Pd and Pt only molecular adsorption is observed and this is completely reversible
(55). The chemisorption pressure however must not be too high (2-20 Torr), in order to avoid the
formation of carbonate species, that would alter the measurement results (56).
21
2.3 Powder X-Ray Diffraction
X-ray crystallography is a well established technique to explore the structure of a material.
The method is based on the diffraction phenomena occurring when X-ray photons are scattered by
the crystallographic planes of a lattice, and revolves in particular on the interference patterns given
by the diffracted photons. The intensity of the diffracted beam is maximal when all the diffracted
rays give constructive interference and is described by the Bragg's law:
n (2.12)
w -ray, d is the interplanar
X-ray diffractograms reveal several important properties of a material, namely the
crystallinity of the sample or of a particular component of it, an estimate of the size of the
microcrystallites that may be present, the atomic constituents of the unit cell and so on (47).
Powder XRD was used in the current study to identify the composition of the phases and to
estimate the average dimension of the crystallites. Most of the powder XRD patterns were recorded
with a computer- e
The experimental broadening of the XRD reflections is composed of many contributions,
but it can be related to the size of the crystallites of the studied material, according to the
Scherrer's equation:
(2.13)
where is the crystallites mean size, K is a constant which to some degree depends on the
- Full Width at Half Maximum (FWHM) of the
According to the Scherrer's equation, reflections having larger FWHM coincide with lower
dimension of the crystallites. For the same principle, non-crystalline materials show no sharp
diffraction reflections, but only broad features, because of the absence of a long-range order (47).
However, it should be noticed that the Scherrer equation only provides a lower bound on the
particle size, because the reflections may be broadened by a variety of factors besides crystallite
size. Some examples of reflection broadening sources may be dislocations, stacking faults, residual
stresses, grain boundaries, impurities, etc. If all of these contributions were zero, then the peak
22
width would be determined solely by the crystallite size and the Scherrer formula would apply.
However, if the other contributions to the peak width are non-zero, then the actual crystallite size is
cherrer's
equation is affected also by instrumental effects.
2.4 Fourier Transform Infrared Spectroscopy
Fourier Transform Infrared Spectroscopy (FTIR) is a technique used to acquire absorption or
emission infrared spectra of a sample, that relies on a Fourier transform mathematical process to
convert the raw data (interferogram) into the actual spectrum. Contrary to dispersive spectroscopy
techniques, in which a monochromatic light beam is used to measure how much of the light is
absorbed at every wavelength, FTIR allows to simultaneously collect high spectral resolution data
over a wide spectral range.
An FTIR spectrometer consists of a polychromatic IR source, a Michelson interferometer
allowing to block or transmit a certain wavelength of light, a sample holder/chamber and a
detector. The most common IR source for the mid- 400 cm 1), is a silicon
carbide element heated to about 1200 K, giving an output which is similar to a blackbody. The light
coming from the source is collimated and directed to a beam splitter, usually made of KBr with a
germanium-based coating that makes it semi-reflective. Here, some of the light is refracted
towards a fixed mirror and some is transmitted towards a moving mirror. Light is reflected from the
two mirrors back to the beam splitter and some fraction of the original light passes into the sample
compartment. There, the light is focused on the sample and finally refocused on to the detector.
The difference in optical path length between the two arms of the interferometer is known as the
retardation, or optical path difference, OPD. The interferogram is obtained by varying the OPD and
recording the signal from the detector for various values of OPD. Mid-IR spectrometers commonly
use pyroelectric detectors that respond to changes in temperature due to IR radiation intensity
variation (e.g. deuterated triglycine sulfate (DTGS) or lithium tantalate (LiTaO3)). These detectors
operate at ambient temperatures and provide adequate sensitivity for most routine applications.
The three principal advantages of FTIR compared to scanning dispersive IR are higher
signal-to-noise ratio for a given scan time, higher wavelength accuracy and less sensitivity to stray
light, which is radiation of one wavelength appearing at another wavelength in the spectrum. The
higher signal-to-noise ratio arises from the fact that information from all wavelengths is collected
simultaneously and that no slits are needed: the interferometer throughput is determined only by
the diameter of the collimated beam coming from the source. The higher wavelength accuracy is
due to the fact that the scale is calibrated by a laser beam of known wavelength that passes
23
through the interferometer. This is much more stable and accurate than in dispersive instruments
where the scale depends on the mechanical movement of diffraction gratings.
In this study, DRIFTS spectra were acquired using a Mattson Galaxy 2020 FTIR spectrometer
with a diffuse-reflectance attachment (Collector II) purchased from Spectra-Tech Inc., using a
resolution of 16 cm-1. Since the sample could not be heated to sufficiently high temperatures in the
sample stage of the diffuse reflectance unit, the catalysts were first treated under different
conditions in a separate flow reactor, cooled to room temperature in He, and then transferred to a
sample holder for the spectroscopic measurements. In the diffused reflectance cell, the catalyst
was degassed in dry flowing He at 200 °C before data collection. All spectra were recorded at room
temperature in dry He.
2.5 Transmission Electron Microscopy Techniques
In the field of heterogeneous catalysis and photocatalysis, the design of more active and
selective catalysts often relies on the precise identification of active sites. Sophisticated imaging
methods that enable detailed characterization of a sample at the nanometer and atomic level are
of pivotal importance (57). Among many characterization techniques, advanced electron
microscopy techniques are the most powerful to get information on the individual components of
an heterogeneous material. These techniques are essential to understand the properties of
heterogeneous catalysts and to provide useful information for the development of nanostructured
materials. Using modern electron microscopes it is possible to directly observe small particles,
clusters or even single atoms of a sample, while all other techniques (e.g. X-ray techniques, IR
spectroscopy, NMR spectroscopy) provide information averaged over millions to trillions of
components, or they require stringent conditions on the samples to be examined (e.g. Scanning
Probe Microscopy techniques) (57).
Transmission Electron Microscopy (TEM) techniques overcome the limitation of light
microscopes for imaging very small objects, thanks to their higher resolution. The resolution of a
microscope is the minimum distance between distinguishable objects in an image and it's limited
by two different and unrelated aspects: aberration and diffraction. Aberration can be explained by
geometrical optics and can (in principle) be solved by increasing the optical quality of the system.
Diffraction is instead strictly related to the nature of the wave used for the observation.
The maximum theoretical resolution of a microscope, limited by diffraction alone, is related
to the wavelength of the radiation used according to the Rayleigh criterion, that leads to the
simplified statement that the limit of resolution of any imaging process is on the order of the
wavelength of the wave used to image it. It follows that the maximum theoretical resolution of a
24
good light microscope (hundreds of nm) is not useful to characterize the typical nanocomponents
of a heterogeneous catalyst. On the other hand, the maximum resolution for an electron
microscope is adequate to get nanoscopic resolution, and can be calculated referring to de
Broglie's wavelength of the particle:
(2.14)
where p is the particle momentum wavelength and h is the Planck's
constant.
In the TEM apparatus the electrons are accelerated by a potential drop, V, acquiring a
potential energy eV that is converted to kinetic energy of the electrons at the end of the
accelerating section of the instrument. Equaling the two energies, an expression of momentum as
a function of the potential is derived:
(2.15)
(2.16)
The value tained substituting Equation 2.16 in Equation 2.14. For example, an
electron accelerated to 100 keV has a wavelength (which more or less corresponds to the
maximum theoretical resolution of a microscope) of about 0.004 nm, which is 100 times smaller
than the diameter of an atom. Moreover, by increasing the accelerating voltage the wavelength of
the electrons would decrease. However, it must be pointed out that Equations 2.14 - 2.16 do not
take into account relativistic effects, that cannot be ignored for energies above 100 keV. In addition
to this intrinsic limitation, considerable practical limitations involved in the microscope
construction must be taken into account, such as aberration limits and non homogeneity of the
magnetic fields used as lenses. Nonetheless, atomic scale resolution is attainable using modern
instruments. Using aberration correctors (usually referred to as Cs-TEM), it is possible to
dramatically improve the spatial and spectral resolution of the electron microscope even when
using lower accelerating voltages (Figure 2.3) (57). This is particularly advantageous since a lower
accelerating voltage generally leads to less sample damage from the electron beam.
25
Figure 2.3
Spatial resolution versus year for optical microscopes and electron
microscopes. Currently, best point-to-point spatial resolution is 0.5 Å.
(Reproduced from (57))
A typical Transmission Electron Microscope (TEM) consists of a vertical column in which the
electron beam passes from an electron source at the top, through the specimen and down to the
bottom of the column where the image is formed and revealed (Figure2.4). The column is held
under ultra-high vacuum (UHV) conditions by a system of high performance pumps in order to
reduce the scattering of the electron beam by gas atoms. Electromagnetic coils that function as
lenses are positioned around the column along its length and work in an analogous way the optical
lenses do in a light microscope. Apertures of different diameters can be inserted into the electron
beam at several positions along the column. This is done to select part of the beam and exclude
the contribution of the rest. There are two common types of electron sources, which are
characterized by the way in which the electron beam is generated (thermionic emission source or
field emission gun).
After electrons are produced, they are focused and accelerated by an electrostatic field and
they enter into the TEM column. The strength of this field determines the kinetic energy of the
26
electron beam (Equation 2.14 and 2.16). In practice, an applied potential of at least 100 kV is
advisable for HRTEM.
Figure 2.4 Schematic representation of the TEM column. Blue lines represent the electron
beams.
The condenser apertures and lens system select the electrons and focus them in a parallel
beam, setting the beam diameter to a desired value. Then, the electron beam passes through the
specimen, which is mounted on the specimen holder. The incident beam interact with the internal
crystal structure of the sample and emerges as a set of diffracted and non-diffracted (transmitted)
beams. These beams are again focused by the objective lens on the Back Focal Plane (BFP). In this
plane, the diffraction pattern is formed.
Considering the rules for diffraction (similarly to single crystal X-ray diffraction) each spot is
related to one family of crystal planes at one particular orientation. The objective lens performs a
Fourier Transform on the exit wave to form the diffraction pattern, that is processed via inverse
Fourier Transform to construct an image closely related to the internal structure of the sample.
27
Images and diffraction patterns are visualized on a fluorescent screen and may be digitally
recorded using a camera positioned below the fluorescent screen (58).
The diffraction pattern contains electrons from the whole area of the specimen illuminated
by the beam and is not very useful because the specimen will often be buckled and because the
direct beam is often so intense that it will damage the camera. Therefore, there are basic TEM
operations that allow both to select a specific area of the specimen to contribute to the diffraction
pattern and to reduce the intensity of the diffraction pattern reaching the screen. There are two
ways to perform this operation: either the beam is made smaller, or an aperture is inserted above
the specimen so that only electrons passing through it may hit the specimen. Usually the second
way is applied: this operation is called Selected-Area Diffraction (SAD).
After the diffraction pattern area is selected, it is possible to perform the two most basic
imaging operations in the TEM. It is possible to form the image in the TEM by using the central spot
or some of the scattered electrons. If the direct beam is selected, the resultant image is called
bright field (BF) image, while if scattered electrons of any form are selected, the resultant image is
called dark-field (DF) image. The BF detector is aligned to the transmitted beam in the TEM column
while the DF detector is usually annular and surrounds the BF detector. An image obtained with
the annular detector is called Annular Dark-Field (ADF) image. The DF detector will collect not just
scattered electrons but also some Bragg electrons. To get an image formed only by scattered
electrons another detector can be used, placed at very high angles, called High-Angle ADF
(HAADF) detector. The images collected with this detector are sometimes called Z-contrast images
because the signal strength of the high-angle scattered electrons is proportional to approximately
Z2 (in the limit of electron single scattering). This technique is particularly advantageous for
detecting supported heavy-metal particles. The complete HAADF detector set-up for Z-contrast
imaging in a STEM is depicted in Figure 2.5.
28
Figure 2.5
Schematic representation of the HAADF detector set-up for Z-contrast imaging
in a STEM. The conventional range of electron scattering angles gathered by
each detector is also reported.
HRTEM is a very powerful technique for the characterization of crystallinity, defect
structure, symmetry, morphology and facets of individual nanoparticles. However, it presents some
limitations which must be remembered. Since it is a transmission technique, the samples must be
electron transparent and, in addition, must be thin enough to minimize multiple diffraction of the
electrons. Generally the specimens must have a thickness below 100 nm, or in some cases below 50
nm. Moreover, the high energy electron beam may alter and damage the specimen by heating it or
by reducing susceptible materials such as some oxides. Nonetheless, it must be taken into account
that HRTEM images are 2-D representation of a 3-D set of objects viewed by transmission rather
than reflection, and sometimes the third dimension cannot be neglected. Recent developments in
electron tomography (ET) and quantitative scanning TEM (STEM) have further expanded the
abilities of TEM by allowing 3-dimensional (3D) reconstructions of nanomaterials instead of only 2-
dimensional (2D) projections (59).
STEM operation mode is another way to perform the analysis of the specimen, consisting in
scanning a focused convergent beam over the sample using a system of additional lenses, usually
incorporated in the instrument. This operation mode has the advantage that the sample doesn't
29
have to be moved to get more than one image. Moreover, less noisy images than TEM DF can be
obtained via the combination of the STEM mode with the ADF detector. The ability to quantify the
size of individual nanostructures in the STEM is primarily determined by the size of the electron
beam and the stability of the nanoclusters under the intense electron irradiation of the beam. The
recent development of aberration correctors for STEM has largely improved the spatial resolution
allowed by these instruments, taking the resolution to less than 0.1nm. However, this advance does
not automatically improve sensitivity/accuracy when the size measurement is limited by electron
irradiation effects such as chemical changes and the movement of the nanoclusters on the support
surface.
Inelastic interactions of the beam electrons with the specimen, in which energy is
transferred, also provide useful complimentary information and several analytical techniques have
grown up around the TEM to exploit this. Energy transfer from an incoming high energy electron to
one of the core electrons of an atom may result in ionization of the atom, forming a hole in the
core. An electron in a higher energy level of the atom may then recombine with this hole, releasing
its excess energy as an X-ray photon. The frequency of these photons is determined by the
difference between the two electron energy levels and will therefore be characteristic of the nature
of the atom from which it was emitted. In X-ray Energy Dispersive Spectroscopy (XEDS or EDX),
these X-ray photons are collected and number of counts plotted against their energy to give a
spectrum. This technique allows quantitative measurement of the elemental composition of the
area of the sample illuminated by the beam and provides information which is not available from
HRTEM itself. For this reason, XEDS spectrometer is commonly fitted to the modern TEM.
2.6 X-Ray Absorption Spectroscopy
X-ray Absorption Spectroscopy (XAS) is a well-established technique used for determining
the local geometric and/or electronic structure of matter (60). XAS measures the energy-dependent
fine structure of the X-ray absorption coefficient near the absorption edge of a particular element.
-rays of intensity I0 are incident on a sample, the extent of absorption
depends on the photon energy E and sample thickness t.
(2.17)
where It -dependent X-ray absorption coefficient.
30
(2.18)
where d is the target density and Z and m are the atomic number and mass, respectively (61).
he photon energy equals or
exceeds the binding energy of a core electron, another core electron can be excited and a sharp
increase in absorption coefficient is observed. Above the absorption edge, the difference between
the photon energy and the binding energy is converted into kinetic energy of the photoelectron
ergy. Core-hole states have a short
lifetime (10 15 s), after which an electron from a higher energy state fills the hole, releasing the
energy difference via fluorescence X-ray or Auger electron emission.
According to quantum mechanical perturbation theory, the transition rate between the
core level and the final state is proportional to the product of the squared modulus of the matrix
eleme :
(2.19)
where |i⟩ and |f⟩ are the initial and final state, respectively, and Hp is the electromagnetic field of
the X-ray photon, the Hamiltonian that causes the transition (60). Both factors can cause a
modulation of the absorption coefficient thus creating the X-ray absorption fine structure (XAFS).
At the smallest X-ray energies for which the photon can be absorbed, the photoelectron will be
excited to unoccupied bound states of the absorbing atom. This can lead to a strong increase of
the absorption coefficient at particular X-ray energies corresponding to the energy difference
between the core level and the unoccupied states (pre-edge absorption bands). For higher X-ray
energies, the photoelectron is promoted to a free or continuum state. The wave thus created
propagates outwards and is scattered at neighboring atoms. The outgoing and scattered waves
interfere in a manner that depends on the geometry of the absorber environment and on the
photoelectron wavelength, which is inversely proportional to the photoelectron momentum and
therefore changes with photon energy. Thus, the final state is an energy-dependent superposition
of outgoing and scattered waves. Because the initial state is highly localized at the absorbing atom,
the matrix element M in (2.19) depends on the magnitude of the final state wave function at the
site of the absorbing atom. Constructive or destructive interference of outgoing and scattered
31
waves thus increases or decreases the absorption probability, creating an energy-dependent fine
structure of the absorption coefficient. Figure 2.6 schematically shows the absorption fine structure
as a function of photon energy. Three regions are commonly distinguished: the pre-edge region,
the X-ray absorption near edge structure (XANES) and the extended X-ray absorption fine structure
(EXAFS)(62).
Figure 2.6 Example of XAS spectrum with the three regions of pre-edge, XANES and
EXAFS highlighted.
XANES is characterized by transitions of the photoelectron to unoccupied bound states and
is therefore sensitive to the chemical bonding, exhibiting characteristic features for different
oxidation states and compounds of the absorbing atom(60). The XANES features are also
influenced by strong multiple scattering effects which depend on the three-dimensional geometry
of the local structure around the absorbing atom. This provides a means of distinguishing between
different crystal phases. Although significant progress has been made over recent years, theoretical
calculations of the fine structure in this region are complex and the accuracy of such simulations is
still limited. Therefore, the measured spectra are usually compared to those of known standards
32
and the ratios in which the standards are present in the sample are calculated using linear
combination fitting.
In the EXAFS region (photon energies higher than ~30 eV above the edge), the
photoelectron is promoted to a free or continuum state, and the signal depends on the atomic
arrangement around the absorber (60). EXAFS contains information about the coordination
number, interatomic distances and structural and thermal disorder around a particular atomic
species. An advantage of EXAFS with respect to XRD is that it does not require long-range order
and is applicable to a wide range of ordered and disordered materials. Theoretical calculations of
the fine structure in the EXAFS region have also improved enormously during the last two decades
and simulations with sufficient accuracy are now available. Nevertheless, the measurement of
suitable standards still constitutes an important part of the experimental procedure.
Most XAS experiments are performed at synchrotron sources due to the requirement of
high X-ray intensities and a continuous energy spectrum. In a typical XAS beamline, mirrors are
used to collimate and focus the beam while apertures and slits define its size. In a classical XAFS
beamline, a double crystal monochromator is used to select X-rays of a very narrow energy band
using the criterion for Bragg diffraction. Harmonic energies (that satisfy the Bragg condition with n
) are removed from the beam by slightly detuning the monochromator, which decreases the
transmission of harmonics significantly more than that of the primary energy. Alternatively, X-ray
mirrors can be used that only reflect energies below a critical value. With such an experimental
arrangement, the absorption coefficient can be measured as a function of X-ray energy.
In general, the absorption coefficient can be detected either directly by measuring the
intensities of incoming and transmitted beam (transmission mode) or indirectly by measuring the
intensity of the incoming beam and of the decay products, such as fluorescent X-rays or Auger
electrons (fluorescence or electron yield mode). The experimental setups for the various
acquisition modes during XAS experiments are shown schematically in Figure 2.7.
33
Figure 2.7 Schematic representation of the experimental setup for the different XAS
detection modes
Once the signal is recorded, XAS data require some mathematical analysis in order to be
meaningful. There exists a variety of ways to analyze XAS data and a large number of codes and
programs are available. In general, data from the XANES and EXAFS region are analyzed separately.
This is due to the different information contained in both spectral regions and to the fact that
theoretical modeling for XANES is not yet as advanced as it is for EXAFS. The calibration of the
energy scale and the alignment of different spectra is very important in both cases: in the XANES
for the determination of edge shifts and in the EXAFS for accurate bond length determination. For
each absorption edge, the beamline monochromator has to be calibrated with a known reference,
often a thin metal foil, measured simultaneously with each sample of interest.
The first step in XANES analysis is the normalization of the spectrum: this removes effects of
sample thickness and concentration and allows the direct comparison of different samples and
measurements. The spectrum in the energy region before the absorption edge is fitted by a linear
pre-edge line, while the spectrum well above the absorption edge is typically approximated by a
quadratic post-edge line. The edge step Δ 0 is obtained as the difference between pre-edge and
post-edge lines at the absorption threshold E0. A normalized XANES spectrum is finally obtained by
subtracting the pre-edge line from the measured spectra over the whole energy range and
dividing by Δ 0, therefore flattening the spectrum after the threshold. The normalized spectrum
34
can be compared with spectra of standard compounds. The relative amount of the most important
components of the sample can be determined by fitting the experimental normalized spectrum
with a linear combination of the spectra of standard compounds, the so-called Principal
Component Analysis (PCA).
EXAFS analysis requires more steps: first, the fine structure Χ(E) is isolated from the
absorption background, fitting the pre- and post-edge lines. The pre-edge line is subtracted from
the spectrum over the whole energy range and t 0(E) is typically
approximated by a spline function that approaches the post-edge line at energies well beyond the
absorption edge. 0(E)
is normalized with respect to the step height Δ 0 , yielding the fine structure Χ(E). In order to
calculate Χ(k), the threshold energy E0 is needed: this is typically taken as the maximum of the
the half of the step height.
The second step is a Fourier transform of the data into R-space. Fourier transformation (FT) of the
EXAFS provides a means to visualize different scattering contributions and is often used during
analysis. When Fourier transformed, different scattering contributions with a large difference in Rj
will ,
one should remember that the FT is a complex function and both magnitude and phase (or
alternatively, real and imaginary part) have to be considered for the full information content.
A back-transformation can be used to isolate different scattering contributions if their
signals are well separated in R-space. However, this procedure usually fails for higher coordination
shells due to the overlap of different scattering contributions. To overcome this problem, the path
fitting approach can be used. Path-fitting is a model-dependent approach based on the cumulant
expansion of the different single and multiple scattering paths and requires some pre-existing
knowledge about the system under investigation. The analysis starts with a model structure that
specifies the absorbing atom and the position and type of the surrounding atoms that are to be
considered in the fitting procedure. The biggest advantage of the path fitting approach is the
ability to analyze the structural parameters beyond the first coordination shell which often contain
crucial information not available from the first nearest neighbor environment.
EXAFS experiments on Pd@CeO2-based catalysts for methane oxidation (Chapter 3) were
performed at the SAMBA beamline of Synchrotron SOLEIL (France) with a Si 220 double crystal
monochromator. The monochromator was kept fully tuned and harmonics were rejected by a pair
of Pd-coated, Si mirrors. Spectra were measured in transmission mode using ionization chambers
as detectors. One chamber was used as the baseline monitor and two other chambers were used to
continuously check the stability of the energy scale by placing one after the sample and the other
35
after a reference foil. Because we used Pd mirrors to reject harmonics, the baseline was checked to
ensure the absence of a residual Pd signal due to nonlinearity in the detectors. The experiments
were conducted using the transmission and fluorescence cell fully described in (63). 8 mg of
catalyst, diluted 1 : 10 by weight with BN, and a total gas flow rate of 12 mL min-1 (0.5 % CH4, 2.0 %
O2 , 15% water if needed, and N2 balance) were used for Ce L(III) edge experiments. 95 mg of
catalyst and a total gas flow rate of 130 mL min-1 (0.5 % CH4, 2.0 % O2 , 15% water if needed, and N2
balance) were used for Pd K edge experiments. Products analysis systems was a Cirrus - MKS mass
spectrometer, and activity comparable to that observed during catalytic test were observed. The
fractions of Ce(III) and Pd(0) in the samples were determined by fitting the XANES (X-Ray
Absorption Near Edge Spectroscopy) part of the spectrum using a linear combination of spectra for
CeAlO3 and CeO2 to fit data for Ce and a combination of spectra for PdO and a Pd foil to fit data for
Pd. All measurements were performed during the same beam session. EXAFS data were fit with
Feff8.4 (64) and Horae (65) software packages. Fits were performed in r-space and a Hanning
window was applied in k-space from 3 to 13 A-1 (the window was zero outside these values).
EXAFS experiments on Pt-Co based catalysts for hydrodeoxygenation of biomass-derived
feedstock (Chapter 4.3) were performed at the beamline 5BM-D at the Advanced Photon Source,
Argonne National Laboratory. Catalyst samples were diluted with boron nitride and pressed into
pellets in a six-well sample holder. The catalysts were reduced at 250 oC (pre-treatment
temperature) and 400 oC (higher than the pretreatment temperature) for 1 h (ramp rate 5 oC min-1)
under atmospheric pressure in a 40 ml min-1 hydrogen flow prior to measurement at ambient
temperature under hydrogen. Data processing was done using the program Athena of the
Demeter suite. Fitting of the EXAFS oscillations was done using the Artemis program of the same
suite. For the Pt L3 edge, oscillations were fitted using the Fourier transform (FT) from
wavenumbers k = 3 Å-1 to k = 12 Å-1 and R = 1.6 Å to 3.3 Å while for the Co K edge, oscillations were
fitted using a FT from wavenumbers k = 3 Å-1 to k = 12 Å-1, and R = 1.4 Å to R = 3.2 Å. Amplitude
reduction factors were calculated using the EXAFS spectrum of the Pt or Co foils and assuming a
coordination number (CN) equal to 12. A single Debye-Waller factor was used for all scattering
paths.
2.7 X-Ray Photoemission Spectroscopies
Photoemission spectroscopy (PES) refers to the measurement of the kinetic energy and
number of photoelectrons, i.e. electron emitted from a sample by the photoelectric effect, in order
to determine the oxidation state of the chemical species (through the binding energies of
electrons) and their relative amount in the sample under investigation. Since photoelectrons have
a short inelastic mean free path in matter, PES techniques are very sensitive to the composition of
the sample surface (in the order of 1-10 nanometers), and are thus very relevant for studying
36
catalytic materials (66). PES is a general term indicating various techniques, depending, among
others, on the photon energy (Ultraviolet Photoemission Spectroscopy UPS; X-Ray
Photoemission Spectroscopy XPS), on the source of photon (Synchrotron Radiation
Photoelectron Spectroscopy SRPES) and the pressure in the sample chamber (Near-Ambient
Pressure and Ambient Pressure XPS NAP-XPS and AP-XPS).
The power of XPS for chemical analysis derives from the fact that electron core levels have
small chemical shifts depending on the chemical environment of the atom that is ionized. XPS
allows to determine the formal oxidation state of the atom, the identity of its nearest-neighbors
and its bonding hybridization to that nearest-neighbor atoms. XPS is routinely used to analyze
inorganic compounds, metal alloys, semiconductors, catalysts, ceramics and many others. It is a
powerful tool to detect and quantify impurities, since the detection limits for most of the elements
(on a modern instrument) are in the parts per thousand range. One limit of XPS is that it requires
high vacuum (P ~ 10 8 millibar) or ultra-high vacuum (UHV; P < 10 9 millibar) conditions, strongly
limiting XPS use for insightful in-situ analysis of catalysts. However, a current area of development
is adopting differential pumping in order to expose the sample to higher pressure (tens of millibar
for NAP-XPS) and overcoming this obstacle. Another intrinsic drawback of XPS is that some
materials can be sensitive to X-rays or vacuum degradation, causing a modification of the sample
upon analysis.
XPS can be performed using a -
rays source, or using a synchrotron-based light source combined with a custom-designed electron
energy analyzer (SRPES). Since the energy of an X-ray with particular wavelength is known (1486.7
eV and the emitted electrons' kinetic energies are measured, the electron binding
energy of each of the emitted electrons can be determined by using the energy conservation
equation:
(2.20)
where BE is the binding energy of the electron, Ep is the energy of the X-ray photons, Ek is
the kinetic energy of the photoelectron as measured by the instrument and is the work function,
an adjustable instrumental correction factor, dependent on both the spectrometer and the sample.
It accounts for the few eV of kinetic energy given up by the photoelectron as it becomes absorbed
by the instrument's detector. In practice, it is a constant parameter that rarely needs to be adjusted.
A typical XPS spectrum is a plot of the number of electrons detected versus their BE. Each
element produces a characteristic set of XPS peaks at characteristic BE values, corresponding to the
37
electron configuration of the electrons within the atoms. The number of detected electrons for a
certain signal is directly related to the amount of atoms of element within the XPS sampling
volume, so atomic percentages can be determined, upon correction by a relative sensitivity factor
and normalization. The quantitative accuracy depends on several factors, such as signal-to-noise
ratio, peak intensity, surface volume homogeneity, correction for energy dependence of electron
mean free path and degree of sample degradation due to analysis.
The main components of a XPS system (Figure 2.8) include a source of X-rays, an ultra-high
vacuum (UHV) stainless steel chamber with UHV pumps, an electron collection lens, an electron
energy analyzer, magnetic field shielding, an electron detector system (electron multiplier), a
moderate vacuum chamber for sample introduction, sample mounts and stage, and a set of stage
manipulators.
Figure 2.8 Graphical representation of a XPS system.
The development of the near ambient pressure XPS (NAP-XPS) in the past decades made it
possible to investigate surfaces under reactive environments. The main experimental problem is
38
the electron mean free path in the gas phase (e.g. 0.04 cm at 1 mbar for 50 eV electrons or 0.4 cm at
1000 eV), which is much shorter than the classic sample-analyzer distance (15 mm), making the
probability to detect photoelectrons extremely low. This problem has been overcome by placing
the sample surface very near the electron energy analyzer entrance and by differentially pumping
the analyzer. To do so, the analyzer entrance diameter has to be small, which results in a small
sampled area. A small focal spot of X-rays (~100 μm) impinging on the sample is also
indispensable.
In the present work, NAP-XPS experiments were carried out in a custom-built system
(SPECS Surface Nano Analysis, GmbH Germany) equipped with a PHOIBOS 150 Hemispherical
Energy Analyzer, coupled with a differentially pumped electrostatic pre-lens system. The reaction
NAP cell was installed in an analysis chamber that had a base pressure of ~10-10 mbar and allowed
in situ XPS studies at pressures up to 20 mbar. A high-intensity, monochroma -ray
source (1486.6 eV) was used to record the spectra of Cu 2p, Ni 2p, O 1s, P 2p, N 1s and C 1s core
levels. Binding energies are reported after correction for charging, using metallic Cu 2p3/2 signal
(932.4 eV) as a reference.
2.8 Atomic Force Microscopy
Atomic Force Microscopy (AFM) is a type of scanning probe microscopy, with
demonstrated resolution on the order of fractions of a nanometer. The AFM has three major
abilities: force measurement, imaging, and manipulation. For imaging, a mechanical probe (tip)
attached to a cantilever is scanned on the specimen surface, so that interatomic forces between
the tip and the specimen surface lead to a deflection of the cantilever, according to Hooke's law
(Figure 2.9). Forces that are measured in AFM include mechanical contact force, van der Waals
forces, capillary forces, chemical bonding, electrostatic forces, magnetic forces, etc. The deflection
of the cantilever is detected (e.g. by interferometry, piezoresistive methods) and converted to an
electrical signal of intensity proportional to the deflection, so that a trace of the sample surface is
outlined during a scan. In practice, several different imaging modes can be used: either the value of
the deflection (contact mode), or the amplitude of an imposed oscillation of the cantilever (tapping
mode), or the shift in resonance frequency of the cantilever (non-contact mode) can be monitored.
Raster scanning of the sample gives a tridimensional map of the surface, commonly displayed as a
pseudo-color plot.
39
Figure 2.9 Graphical representation of an AFM system.
Compared to competitive technologies such as optical microscopy and electron
microscopy, the AFM does not require lenses or beam irradiation, avoids loss of space resolution
due to diffraction limit and aberration, and does not require vacuum to operate. Unlike the SEM
electron microscope, which provides a two-dimensional image of a sample, the AFM gives a three-
dimensional surface profile. In addition, AFM does not require special sample treatments such as
metal/carbon coating, usually employed in SEM for insulating materials.
Disadvantages of AFM are the single scan image size and the scanning speed. In one pass,
AFM can only image an area up to 150×150 m, so a typical scan requires several minutes, while
SEM can image an area up to square millimeters, allowing to scan at near real-time, although at
relatively low quality. Parallel probes may be used to improve the size of the scanned area in AFM,
but this is not always feasible. AFM images can also be affected by nonlinearity, hysteresis, creep of
the piezoelectric material and image artifacts. Some artifacts are unavoidable, and could be
induced by an unsuitable tip, a poor operating environment, or even by the sample itself (presence
of steep walls or high-curvature features).
2.9 Scanning Electron Microscopy
Scanning electron microscopy (SEM) is a type of electron microscopy that produces images
of a sample by scanning it with a focused beam of electrons. The beam interaction with the sample
produces a variety of signals containing information about the sample's surface topography and
40
composition. Secondary electrons emitted by atoms excited by the electron beam are most
commonly analyzed (Figure 2.10). The number of secondary electrons that can be detected
depends, among other things, on specimen topography. Resolution better than 1 nanometer can
be achieved. Since the electron beam is very narrow, SEM micrographs have a large depth of field,
yielding a three-dimensional appearance useful for understanding the surface structure of a
sample. Specimens are usually observed in high vacuum, but modern applications allow low
vacuum and wet conditions (in environmental SEM ESEM) operations.
Figure 2.10
The types of signals produced by an SEM and the information they contain.
Secondary electron detectors are standard equipment in all SEMs, but it is rare
that a single machine would have detectors for all other possible signals.
Interaction volume is depicted by a blue drop.
The different signals result from interactions of the electron beam with atoms at various
depths within the sample. The primary electrons (of the beam) lose energy by repeated random
scattering and absorption within a teardrop-shaped volume of the specimen known as the
interaction volume. The interaction volume extends from less than 100 nm to 5 µm into the
surface, depending on the electron's landing energy, the atomic number of the specimen and the
41
specimen's density. The secondary electrons are emitted from very close to the specimen surface,
while back-scattered electrons (BSE) and characteristic X-rays (EDX) emerge from deeper locations
within the specimen. BSE and EDX are used in analytical SEM, because they are dependent on the
atomic number (Z) of the specimen, providing information about the distribution of different
elements in the sample.
For conventional imaging in the SEM, specimens must be electrically conductive, at least at
the surface, and electrically grounded to prevent the accumulation of electrostatic charge at the
surface, scanning faults and image artifacts. Insulators are therefore usually coated with an
ultrathin coating of electrically conducting material (such as Au, Pt, Ir, W, Cr or graphite), deposited
on the sample either by low-vacuum sputter coating or by high-vacuum evaporation. Otherwise,
low-voltage mode SEM operation or ESEM can be employed to reduce charging problems. In low-
voltage SEM operating conditions must be adjusted such that the incoming beam current is equal
to sum of secondary and backscattered electrons currents. In ESEM instruments the specimen is
placed in a relatively high-pressure chamber and differential pumping is used to keep vacuum
adequately low at the electron gun. The high-pressure region around the sample in the ESEM helps
neutralizing charges.
In a typical SEM, an electron beam (0.2 40 keV energy) is emitted from an electron gun (a
heated tungsten filament or a field emission gun). The electron beam is focused by condenser
lenses to a spot about 0.4 5 nm in diameter, and finally deflected by scanning coils or deflector
plates, in order to raster scan the sample surface. Magnification in an SEM can be controlled over a
range of about 6 orders of magnitude from about 10 to 500'000 times. Magnification results from
the ratio of the dimensions of the raster on the specimen and the raster on the display device. It is
therefore controlled by the scanning coils, the deflector plates, and not by objective lens power as
in optical and transmission electron microscopes. Depending on the instrument, the resolution can
fall somewhere between less than 1 nm and 20 nm. The spatial resolution in SEM is limited by the
size of the interaction volume and by the size of the electron spot, which in turn depends on the
wavelength of the electrons and the optical system. The spot size and the interaction volume are
both large compared to the distances between atoms, so the resolution of the SEM is not high
enough to image individual atoms, as is possible transmission electron microscope (TEM). The
advantages of SEM include large area imaging, bulk materials imaging (not possible with TEM), and
the ability to measure the composition and properties of the specimen in a variety of analytical
modes.
42
2.10 Catalytic Activity Measurements
The catalytic tests reported in this thesis were performed on three types of reactivity lines,
that are schematically represented in Figure 2.11-Figure 2.13.
2.10.1 Low Pressure, Gas Flow Reactor Line
For methane catalytic oxidation experiments, the system depicted in Figure 2.11 was used.
Figure 2.11 Schematic representation of the reactivity line used for testing the activity of
catalysts in methane catalytic oxidation.
The gas mixing system allows to pre-treat the catalyst (e.g. reduction or oxidation) and to
simultaneously acquire information on the composition of the reaction mixture, bypassing the
reactor. The mixing system is composed of four flow meters controlled by a central control unit
(Brooks Instruments). A system of valves before the flow meters allows the selection of the desired
gases. A second system of valves positioned after the flow meters diverts each gas to one of the
two following branches: the reaction line or the preparation line. The preparation line is not
43
connected to the analyzer and is therefore usually employed for the pre-treatment of the catalyst.
The reaction line, connected to the analyzer, is used to perform the catalytic tests. A 10 way valve
allows to select which one of the two gas streams passes through the reactor. A saturator is
connected to the reaction line by two switching valves and is used to introduce water vapor in the
reaction gas mixture. All the lines are heated at 125 °C in order to avoid condensation of vapor.
A U-shaped quartz reactor with a 4 mm internal diameter was used for testing powdered
catalysts. The sample was located between two layers of granular quartz in order to sustain the
catalyst powder and preheat the reagents. The reactor was heated by a oven (Micromeritics)
equipped with a PID controller (Eurotherm 847) and the temperature of the catalyst was measured
with a K-type thermocouple8 inserted inside the reactor catalytic bed. In case of model, flat
catalysts, a U-shaped quartz reactor of 1.5 cm internal diameter was used.
The catalytic system was connected to a mass spectrometer Hiden Analytical HPR20,
equipped with a ionization chamber and a quadrupole to separate the ions basing on their mass,
plus two detectors: a Faraday detector and a Secondary Electron Multiplier (SEM) detector. The
catalytic activity were calculated on the basis of the parental ions of the species of interest, after
correction for the cracking pattern of other signals.
2.10.2 High Pressure, Three Phases Reactor Line
For most of the catalytic hydrodeoxygenation reactions, the flow system was designed to
work at high pressures (Figure 2.12). The chosen setup allows to feed H2 gas and a liquid two
desired and independent flow rates. H2 (UHP grade, Airgas) flow rate was controlled by adjusting
the length (corresponding to a certain pressure drop) of a 0.002-inch ID capillary tubing (Valco
Instruments, Inc.). The liquid was injected by a High-Pressure Liquid Chromatography (HPLC) pump
(Series I+, Scientific Systems Inc.), which was also used to measure the total pressure in the reactor.
The pressure within the reactor was controlled by a backpressure regulator (KPB series, Swagelok)
placed at the reactor exit. The liquid flow rates were varied from 0.02 to 0.2 mL min-1, while the H2
flow rates were 2 to 20 mL min-1. The reactor gaseous and liquid effluents were examined
separately by injection into a GC/MS.
Figure 2.12 Schematic representation of the reactivity line used for testing the activity of
catalysts in catalytic hydrodeoxygenation reactions.
8 K-type: non-expensive thermocouple composed of chromel (90% nickel and 10% chromium)- alumel (95% nickel, 2% manganese, 2% aluminium and 1% silicon)
44
A GC-MS (QP-5000, Shimadzu) equipped with a capillary column (HP-Innowax, Agilent
Technologies) was used to measure the products composition. For furfural experiments, the
calibration of liquid samples was accomplished using standard solutions with known
concentrations of furfural, 2-methylfuran (MF), 2-methyltetrahydrofuran (MTHF), furfuryl alcohol
(FA), tetrahydrofurfuryl alcohol (THFA), furfuryl-dipropyl acetal (FAct), furfuryl-propyl ether (FEther),
2-pentanone, 2-pentanol, and pentane. Calibration of the gas-phase samples was verified for furan,
furfural, FA, and MF using the known vapor pressures of the pure compounds. For quantification of
FEther, the GC sensitivity was assumed to be the same as that for FAct. For open-ring pentanedione
and ether products, the GC sensitivity was assumed to be the same as 2-pentanone. Since the
quantities of most of these side products were relatively small, these assumptions will not have a
major impact on the conclusions of the study. Selected HDO experiments were performed in a gas
mixing system similar to the one described in Figure 2.11, saturating the gas feed with furfural
vapor.
For 5-hydroxymethylfurfural (HMF) experiments, the selectivity for each product was
quantified using standard solutions with known concentrations of HMF, dimethylfuran (DMF),
dimethyltetrahydrofuran (DMTHF), 2-hexanone, 2-hexanol, and 2,5-hexandione (all purchased
from Sigma Aldrich). For quantification of other furan-based, intermediate products, the GC
45
sensitivity was assumed to be the same as that for HMF. For open-ring, ether products, the GC
sensitivity was assumed to be the same as 2-hexanone or 2,5-hexandione.
2.10.3 Photocatalytic Reactor Line
The photocatalytic activity of dyes-sensitized Pt/TiO2 was studied under simulated solar
light irradiation in the visible-IR range using a Solar Simulator LOT-Oriel, equipped with a 150 W Xe
lamp, an Atmospheric Edge Filter and a cut-off filter at 420 nm. Irradiance was ~ 6 x 10 3 W m 2 in
the UV-A range and ~ 180 W m 2 in the visible range (400 1000 nm). The photocatalytic system
setup is schematically represented in Figure 2.13.
Figure 2.13 Schematic representation of the photocatalytic reactor used for testing the
activity of dyes-sensitized TiO2.
The reactor is comprised of a cylindrical Teflon-lined stainless steel container connected to
the gas carrier line and hermetically sealed with a quartz window. The catalyst is dispersed with a
magnetic stirrer within the aqueous solution containing the sacrificial agent employed. The reactor
is thermostated at 25 °C. The produced H2 was stripped from the reactor by an Ar flow (15 mL min-1)
and the concentration of H2 in gas stream was quantified using a Agilent 7890 gas chromatograph
equipped with a TCD detector, connected to a Carboxen 1010 column (Supelco, 30 m x 0.53 mm
In order to evaluate and compare the performance of photocatalytic systems, some specific
parameters are used. The Turnover Number (TON) is the number of electrons which react to
produce H2, per active site, before the catalyst becomes inactive. A stable, ideal catalyst would have
an infinite TON. In the specific case of dye-sensitized photocatalytic evolution of H2, the number of
46
electrons that react is equal to 2 times the molecules of produced H2, and the catalytic active sites
may be considered equal to the number of sensitizer molecules, so that the simplified
Equation 2.21 is used in practical experiments:
(2.21)
Since TONs are dependent on the irradiation period, their values should be always referred
to the timescale [e.g., TON(5 h)]. The Turnover Frequency (TOF), that is rate of H2 production per
active site, can also be used to compare photocatalytic activities. It should be noted that TON and
TOF are dependent on many parameters, such as temperature, intensity and wavelength range of
light irradiation.
Apparent Quantum Yield (AQY) and Internal Quantum Yield (IQY) are other parameters that
can be used to compare different catalytic systems. For AQY calculation, the number of catalytic
sites is replaced by the number of incident photons in the TON equation (Equation 2.22). It is called
apparent QY because not all the incident photons are effectively absorbed and/or reach the
reaction center.
(2.22)
For AQY calculation, the number of incident photons is typically measured as a function of
the wavelength by using a monochromatic light source or band-pass filters. In comparative studies,
AQY is usually reported at the same wavelength, even if a more indicative value should be taken at
the maximum of the Vis absorption of the dye-sensitized photocatalyst, which depends on the
specific dye sensitizer.
In the Intrinsic or Internal Quantum Yield (IQY) (Equation 2.23), the absorbed photons
However, IQY is rarely used since the real number of absorbed photons is hard to determine for a
dye/Pt/TiO2 suspension in water, because of light scattering.
(2.23)
47
Solar-to-Hydrogen (STH) energy conversion efficiency is the efficiency of the system in
terms of the amount of incoming solar energy converted into chemical energy in the hydrogen
product (Equation 2.24), where F is the flow of H2 produced (expressed in mol s 1 0 is the
enthalpy associated with H2 combustion (237×103 J mol 1), S is the total incident light irradiance
(expressed in W cm 2), and Airr is the irradiated area (expressed in cm2).
(2.24)
An alternative way to describe the conversion efficiency is the Light-to-Fuel Efficiency (LFE)
(Equation 2.25), in which 0 is the enthalpy associated with H2 combustion (285.8×103 J mol 1).
(2.25)
Both STH and LFE are dependent on the experimental conditions and the irradiation time,
so comparison of STH and LFE values among different studies should be taken carefully. Typical
reported values of STH and LFE are below (in most cases much below) 1%, comparable to the low
efficiency of the natural photosynthesis (67).
48
3. Methane Catalytic Oxidation
3.1 Introduction
Methane is an abundant and available energy source, widely employed for power
generation by thermal combustion and as a fuel in Natural Gas Vehicles (NGVs). However, methane
is a potent greenhouse gas, 20 times as powerful as CO2, so its release in the atmosphere due to
incomplete burning should be mitigated. Catalytic oxidation is widely employed in industrial and
vehicles exhaust after treatments because it drastically reduces the temperature required for
hydrocarbons abatement. Nevertheless, CH4 is the most stable alkane and its activation at low
temperature (<500 °C) is problematic. This represents a serious challenge for environmental
protection and for complying with upcoming emissions regulations. Therefore, better methane-
oxidation catalysts, with high activities at low temperatures and better stability, are needed.
Pd-supported catalysts are the most active materials for methane catalytic oxidation at low
temperature (8, 21, 68 71). Both metallic Pd and PdO are active for methane oxidation, but PdO has
a higher specific activity (8, 72). PdO forms between 300 and 400 °C, and decomposes to metallic
Pd above 800 °C (in air, atmospheric pressure). The decomposition is not a reversible process, so
when PdO is decomposed, its reformation can require a temperature well below 700°C, and this
results in a sharp transient deactivation during cooling of the catalyst. The transition temperatures
depend on O2 pressure and on the interactions between Pd and the support (73). It should be
recognized that the above discussion is a greatly simplified picture of the reaction on working
catalysts because Pd and PdO can co-exist under non-steady conditions and the nature of C-H
bond cleavage changes with the Pd oxidation state (72, 74).
Reducible supports such as CeO2 can act as promoters for Pd oxidation, enhancing stability
and activity of Pd-based catalysts in methane oxidation (75, 76). CeO2 is known to have two
beneficial functions: it provides oxygen for the catalytic oxidation at low temperature by
transferring oxygen to Pd (73, 77, 78) and it can improve the stability of the active PdO phase at
high temperatures (79 81). However, pure ceria has limited thermal stability (82), and direct
contact between the ceria and Pd is required for oxygen transfer.(39, 75) Indeed, recent work has
shown that the activity of ceria-supported Pd catalysts increases with the interfacial contact
between the Pd phase and ceria (39). To maximize this contact, self-assembly methods can be used
to synthesize Pd@CeO2 hierarchical catalysts, which consist of Pd nanoparticles (NPs) surrounded
by a thin porous shell of ceria (8). These catalysts have shown exceptional activity for methane
49
oxidation under dry conditions and increased thermal stability of the catalyst(83). Complete
methane conversion was achieved on a Pd@CeO2/Si-Al2O3 catalyst at temperatures below 400 °C
and at high space velocities. Significantly, the deactivation associated with the classical PdO-Pd
transition observed in conventional Pd-based catalysts was not observed in temperature ramping
experiments, even at very high GHSV (8).
Despite their high intrinsic activity for methane oxidation, Pd-based catalysts suffer from
deactivation under realistic conditions, mainly because of sintering of the metal or support and
because of the presence of poisons in the gas feed, including water vapor, sulfur oxides and
phosphorous compounds (84 91). The aim of this chapter is to report the effect of these common
deactivating agents on the novel Pd-based hierarchical catalysts previously developed by our
research group. Some possible deactivation mechanisms and strategies for enhancing resistance to
poisoning are proposed.
50
3.2 Hierarchical Pd@MOx-based Catalysts Synthesis
The hierarchical catalysts investigated in this chapter are based on core-shell units:
nanocomposites consisted of a noble metal nanoparticle core surrounded by an outer porous
oxide shell, encapsulating it (Figure 3.1). This configuration differs greatly from traditional
supported metal catalysts, that are comprised of small metal particles dispersed on the internal
surface of an oxide support.
Figure 3.1 Schematic representation of core-shell synthesis procedure, starting from MUA
protected metal nanoparticles (Adapted from (92)).
Dispersible core-shell nanostructures having a Pd or Pt core and a TiO2, ZiO2 or CeO2 shell
can be prepared exploiting the versatile self-assembly synthesis previously reported by our group
(92). Here, the supramolecular approach in which single units composed of a Pd core and a CeO2,
ZrO2 or ceria-zirconia (CZ) mixed oxide shell are prepared in solution and then homogeneously
deposited onto hydrophobic supports will be reported. The catalyst preparation requires five steps,
which will be outlined in the following sections.
3.2.1 Synthesis of Functionalized Pd NPs
Materials: Potassium tetrachloropalladate(II) (32.04 % as Pd) was purchased from ChemPur,
11-mercaptoundecanoic acid (MUA, 95 %) from Sigma-Aldrich, Sodium borohydride (98+%) from
Acros Organics, H3PO4 (85% min) from Rectapur and NaOH pellets (>97%) from Carlo Erba Reagenti.
Solvents (reagent grade), were purchased from Sigma- Aldrich and used as received.
The first step in the synthesis of the catalysts studied in the present chapter is the
preparation of Pd NPs functionalized with terminal carboxylic group, which are required for the
reaction with the metal oxide precursor. Briefly, the Pd precursor is reduced in the presence of 11-
51
mercaptoundecanoic acid (MUA), which binds to the growing metal nanoparticle (NP) forming a
three dimensional SAM (Figure 3.2).
Figure 3.2 Schematic representation of MUA-Pd NPs synthesis procedure.
In a typical synthesis, K2PdCl4 (93.63 mg, to get 30 mg of Pd) was dissolved in 10 mL water,
obtaining an orange solution. 50 mL acetone were added and the solution turned red. H3PO4 (1.17
mL, 60 mol vs. Pd) was added, followed by 11-mercaptoundecanoic acid (MUA, 32.40 mg, 0.5 mol
vs. Pd). The solution was stirred for 5 min, then a freshly prepared aqueous solution (4.60 mL) of
NaBH4 (106.64 mg, 10 mol vs. Pd) was rapidly added, causing the solution to turn black
immediately. Stirring was continued for 10 min, and then solvents were evaporated under vacuum
and the particles washed three times with water (20 mL portions) and two times with
dichloromethane (10 mL portions), with sonication and centrifugation (4500 rpm, 15 min) after
each washing cycle. Particles were then recovered by dissolution in THF. The final dispersion was
stable for several weeks.
3.2.2 Cerium (IV) Tetrakis(decyloxide) Synthesis
Materials: cerium ammonium nitrate ((NH4)2Ce(NO3)6, CAN, 99.99%) and sodium methoxide
(ca. 25 wt % in methanol) were purchased from Sigma-Aldrich. 1-Decanol (98+%) was purchased
from Alfa Aesar. The solvents used in the preparation of this compound were stored over activated
3 Å molecular sieves overnight prior to use.
The preparation of cerium(IV) tetrakis(decyloxide) [Ce(C10H21O)4, Ce(ODe)4] followed
previously reported procedures (93, 94). Cerium ammonium nitrate (CAN) (5.00 g, 9.12 mmol) was
dissolved in 50 mL MeOH, after which 1-decanol (6.97 mL, 4 mol vs. Ce) was added. Next, a 25 wt %
solution of MeONa in MeOH (12.51 mL, 6 mol vs. Ce) was introduced drop-wise, causing formation
of gaseous NH3 and precipitation of a bright yellow solid (cerium(IV) methoxide) and a white solid
(NaNO3). The mixture was stirred for 1 h. Then, the solvent was evaporated to yield an orange-
colored oil and NaNO3 precipitate. The oil was dissolved into 25 mL dichloromethane, and the
52
solvent was evaporated again. This procedure was repeated several times, until the precipitate re-
crystallized in bigger grains. Then, the oil was dissolved in DCM and the NaNO3 was filtered out.
Finally, the solvent was removed by evaporation, and the orange-oil product was used without
further purification.
3.2.3 Pd@MOx Core-Shell Units Synthesis
Materials: Zirconium butoxide (80% in 1-butanol), dodecanoic acid (99% minimum) and
THF were purchased from Sigma-Aldrich and used as received.
The hierarchical Pd@MOx(CeO2, ZrO2 and CexZr O2) units were prepared by modifying
published procedures(92). The Pd NPs dispersed in THF were added to a THF solution of
cerium(IV)tetrakis(decyloxide) and/or zirconium butoxide, followed by the addition of a THF
solution of dodecanoic acid. Typically, 10 mL of the THF solution of MUA-Pd nanoparticles (0.5
mg/mL as Pd) were slowly added to 5 mL THF solution of pre-mixed metal alkoxides, keeping
constant the Pd:MOx molar ratio (1:5.5), followed by the addition of dodecanoic acid (1 mol vs. Ce +
Zr) dissolved in 10 mL of THF. The hydrolysis of the metal alkoxide in the Pd Ce/Zr solution was
carried out by slowly adding up to 1.2 mL H2O dissolved in 10 mL THF over 4 h (up to 120 mol vs.
Ce + Zr).
3.2.4 Al2O3-based Supports Preparation and Hydrophobization
Materials were purchased by Sigma-
Aldrich and used as received. High-surface area Al2O3 was purchased by Puralox and calcined at
950°C for 12 hours before use. For model catalysts support preparation, trimethylaluminum (TMA)
was supplied by Cambridge Nanotech, together with the ALD system, while ITO/quartz slides were
purchased from Präzisions Glas & Optik (PGS).
On pristine hydrophilic alumina, hydrophobic core-shell structures agglomerate with one
another rather than adhering to the support (Figure 3.3, part A) (8). The presence of agglomerates
causes loss of accessible active phase surface area, so it was crucial to develop a synthetic way to
modify the support, allowing homogeneous dispersion of the active phase on it. This has been
achieved by reacting alumina with an organosilane, triethoxyoctylsilane (TEOOS): this silane has
three alkoxy groups that are prone to hydrolysis and one alkyl chain that is not, so TEOOS binds to
the support exposing the alkyl chain to the external environment, making alumina surface
hydrophobic. Hydrophobic Al2O3 (Si-Al2O3) exhibits much greater capacity for the adsorption of the
Pd@MOx structures compared to the pristine Al2O3 (Figure 3.3, part B). In the present work, two
different supports were reacted with TEOOS: High-Surface Area (HSA) Al2O3 and Al2O3-modified
ITO/quartz slides, used for model catalysts preparation.
53
Figure 3.3
Schematic representation of hydrophobic alumina preparation, and
comparison of loading of core-shells on pristine and hydrophobic alumina.
(Reproduced from (8)
For HSA Al2O3, 1 g of powder was sonicated in 20 mL of toluene, followed by addition of
TEOOS (0.55 mL). The resulting solution was refluxed for 3 hours and the precipitate powder was
recovered by centrifugation (4500 rpm). The powder was subsequently washed twice with toluene
to remove unreacted TEOOS and byproducts and was dried overnight at 120 °C.
The supports for the model catalysts were prepared by ALD of Al2O3 overlayers of various
thicknesses (2, 5 and 10 nm) on an ITO/quartz support (flat, low-surface-area material). Both TMA
and water precursors were kept at room temperature, resulting in vapor pressures of about 20 Torr
and 11 Torr, respectively. The deposition chamber was maintained at 250°C and a base pressure of
0.1 Torr, with N2 flow of 20 mL min 1. The deposition steps were as follows: 1) pulse water for 0.015
s, 2) hold 5 s, 3) pulse TMA for 0.015 s, and 4) hold 5 s. The thickness of the Al2O3 film increased by
approximately 1 Å/cycle, so that 50 cycles were used to make a 5 nm film. The Al2O3/ITO/quartz
slides were then cut into 9 × 9 mm pieces with a glass cutter and treated in freshly prepared
piranha solution (concentrated H2SO4and 30% H2O2 solution in a 3:1 ratio) in order to clean the
surface from any impurity and increase the hydroxyl group population. The slides were washed
many times with water and acetone prior to being functionalized by reaction in a diluted solution
of TEOOS (0.5 mL) in toluene (20 mL) for 2 days at room temperature.
3.2.5 Impregnation of the Core-Shell Structures on Hydrophobic Supports
For HSA catalysts preparation, the Pd@MOx nanostructures dissolved in THF (50 mL) were
added to the appropriate mass of degassed support to achieve the nominal loading for Pd of 1.0 %
wt. After the mixture was stirred overnight, the solid residue was recovered by centrifugation (4500
54
rpm for 15 minutes) and washed twice with THF. Finally, the powder was dried at 120 °C overnight,
ground to a particle size below 150 m and calcined at 850°C for 5 h, using a heating rate of 3 °C
min-1.
Pd@MOx units were deposited on Si-Al2O3/ITO/quartz model catalysts following a
previously reported procedure (95). Briefly, 0.1 mL of Pd@MOx solution was deposited onto the
support and rinsed with THF after 2 min to remove weakly adsorbed particles. The model catalysts
were finally calcined at 850◦C in air for 5 h, using a heating rate of 3 °C min-1.
55
3.3 Effect of Water
3.3.1 Introduction
Gasoline and lean-burn vehicles exhausts contain large amounts of water vapor (5-15%),
which is detrimental for after treatment catalysts performance (21, 96). It has been reported that
water and methane competitively adsorb on PdO active sites and that water reacts with PdO to
form Pd(OH)2, which is much less reactive with methane (87, 97). There is some disagreement over
whether this deactivation is reversible or irreversible (84, 98, 99), but the inhibition effect of water is
known to be more important at lower temperatures. Burch et al. reported methane-oxidation
reaction rates on Pd/Al2O3 were inhibited by water up to approximately 450 °C (97). The importance
of water inhibition on Pd catalysts also depends on the support, with reducible oxides showing
improved resistance to water inhibition (85). The accumulation of hydroxyl/water on the oxide
support was also suggested to hinder oxygen mobility from the support and lead to deactivation
(100). In this section, the effect of water on Pd@CeO2-based catalysts for methane catalytic
oxidation will be discussed.
3.3.2 Results
The catalytic performance of Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 catalysts for CH4 oxidation
under dry and wet conditions was characterized using light-off curves with a heating and cooling
rate of 10 °C min 1. As shown in Figure 3.4, under dry conditions CH4 conversion began at
approximately 300 °C with increasing temperature and was shifted by approximately 50 degrees to
lower temperatures during the cooling cycle. Subsequent light-off measurements on the same
catalyst were identical to that shown here, demonstrating complete reversibility of the catalyst. As
reported in earlier work (8), no dip in conversion was observed during heating or cooling cycles for
space velocities below 1000 000 mL g 1 h 1. The addition of 15.0 % water to the feed had a dramatic
effect on the light-off curves. Methane conversion was shifted to higher temperature by
approximately 200 °C, and transient deactivations were observed during both heating and cooling
cycles. The decrease in CH4 conversion was small during the upward temperature ramp, down to
98 % of conversion at about 750 °C; but the dip in conversion was very noticeable during cooling,
going down to 60 % conversion at 650 °C.
56
Figure 3.4
Effect of water on methane oxidation light-off curves over Pd(1 wt%)@CeO2(9
wt%)/Si-Al2O3 . Dry conditions (circles): 0.5 % CH4, 2.0 % O2, Ar balance,
O2/O2(stoich)=2, GHSV=200 000 mL g 1 h 1, heating and cooling rates 10 °C min 1.
Wet conditions (squares): same conditions as above, with the addition of
15.0 % H2O.
As previously discussed, transient deactivation for CH4 oxidation is associated with a PdO
→ Pd phase transition (69). When the temperature is ramped upward, conversions can decrease
with increasing temperature, owing to decomposition of PdO, then return to 100 % conversion as
temperature is further increased. The transient deactivation is more evident during the cooling
cycle because the re-oxidation of Pd to the more active PdO is kinetically controlled (101), with Pd
reoxidation occurring through a nucleation mechanism that is favored when some PdO is still
present (72). The transition temperatures depend on the O2 pressure and can be influenced by the
support. The absence of transient deactivations with Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 under dry
conditions is likely caused by the close contact between Pd and ceria in the core-shell
structures. The presence of these transients upon the addition of water to the feed suggests water
suppresses reoxidation of Pd.
Further evidence that the detrimental effect of H2O is due to suppressed oxidation of Pd is
demonstrated in Figure 3.5, which shows light-off curves as a function of the O2 partial pressure.
The magnitude of the dip in the conversion during the cooling cycles decreased steadily with
increasing O2 concentration, while the temperature at which the minimum occurred increased,
57
from 640 °C for 2.0 % O2 °C for 14.0 % O2. The explanation for both of these observations is
that the higher O2 partial pressures help to maintain the Pd as PdO.
Figure 3.5
Effect of oxygen concentration on methane oxidation light-off curves over
Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 under wet reaction conditions: a) 2.0, b) 4.0,
c) 10.0, and d) 14.0 %. Conditions: 0.5 % CH4, different concentrations of O2,
15.0 % H2O, Ar balance, GHSV=200 000 mL g 1 h 1, heating and cooling rates
10 °C min 1.
58
Steady-state measurements of CH4 oxidation over Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 in the
presence of water at 500 and 600 °C were especially revealing. In these experiments, the catalyst
temperature was ramped to the desired temperature in 0.5 % CH4, 2.0 % O2 and 15.0 % H2O and
held at that temperature for a period of time. Then water was removed from the reaction mixture
while monitoring the CH4conversion. Results for experiments performed at 500 °C are shown in
Figure 3.6. After the catalyst temperature reached 500 °C, methane conversion decreased slowly
from an initial value of 90 % to ~60 % after 4 h. When water was removed from the reaction
mixture, the activity was completely restored after less than 10 min. When the sample was cooled
to room temperature at 10 °C min 1, the conversion followed the cool-down curve for dry methane
oxidation reported in Figure 3.4. These results imply that H2O at 500 °C does not modify the
structural properties of the Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 catalyst.
Figure 3.6
Steady state aging test under wet reaction conditions at 500 °C, followed by
activity recovery under dry reaction conditions. Dry conditions (circles): 0.5 %
CH4, 2.0 % O2, Ar balance, O2/O2(stoich)=2, GHSV=200 000 mL g 1 h 1, heating and
cooling rates 10 °C min 1. Wet conditions (squares): same conditions as above,
with the addition of 15.0 % H2O.
59
The results from the analogous experiment at 600 °C (Figure 3.7) were very different. The
CH4 conversion decreased much more sharply during isothermal aging, reaching only 10 % after 4
h. Removing water from the reaction mixture at 600 °C did not completely restore the conversion,
even after 10 h. Finally, when the sample was cooled to room temperature at 10 °C min 1, the
conversion did not follow the cool-down curve for dry methane oxidation reported in Figure 3.4;
but, rather, the conversions were shifted to high temperatures.
Figure 3.7
Steady state aging test under wet reaction conditions at 600 °C, followed by
activity recovery under dry reaction conditions. Dry conditions (circles): 0.5 %
CH4, 2.0 % O2, Ar balance, O2/O2(stoich)=2, GHSV=200 000 mL g 1 h 1, heating and
cooling rates 10 °C min 1. Wet conditions (squares): same conditions as above,
with the addition of 15.0 % H2O.
The catalyst deactivation observed in Figure 3.7 was not caused by a permanent
destruction of the catalyst, as demonstrated by the data in Figure 3.8a. In this experiment, the Pd(1
wt%)@CeO2(9 wt%)/Si-Al2O3 catalyst was again aged in 0.5 % CH4, 2.0 % O2, and 15.0 % H2O for 4 h,
then the sample was purged under Ar and cooled to room temperature. Next, the sample
temperature was ramped to 850 °C at 10 °C min 1. The conversions during this initial temperature
ramp were shifted to significantly higher temperatures than that observed even with wet feed in
Figure 3.4. On the aged catalyst, 100 % conversion of CH4 was not reached until above 650 °C.
However, the cool-down curve followed the conversion data observed for dry CH4 oxidation on the
fresh catalyst and subsequent light-off curves for dry CH4 oxidation were identical to that observed
on the fresh catalyst.
60
Figure 3.8
Methane oxidation light off experiment over a) Pd(1 wt%)@CeO2(9 wt%)/Si-
Al2O3 catalyst and b) reference impregnated Pd(1 wt%)-CeO2(9 wt%)/Si-Al2O3
catalyst, both aged at 600 °C for 4 h under wet reaction conditions and cooled
to 150 °C under Ar. Squares represent first cycles and circles second cycle.
Conditions: 0.5 % CH4, 2.0 % O2, Ar balance, O2/O2(stoich)=2, GHSV=200 000
mL g 1 h 1, heating and cooling rates 10 °C min 1.
61
It is interesting to note that something similar occurs on the reference impregnated
sample. However, aging at 600 °C under wet conditions for 4 hr leads to a deactivation of the
catalyst that is less marked with respect to that of the hierarchical catalyst under the same
experimental conditions (residual conversion ~32 % vs. ~8 %). Upon removing water, a sharp
increase in activity is obtained and complete restoration of the initial activity is achieved after high
temperature treatment. Interestingly, the light-off curve measured under dry conditions for this
aged impregnated catalyst (Figure 3.8b) is once again shifted to higher temperature and the
cooling cycles resembles that of the fresh sample as the following activity.
Simultaneously with the conversion measurements in Figure 3.8a, water evolution was also
monitored over the Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 sample, with results shown in Figure 3.9.
Figure 3.9
Water evolution (signal m/z=18) recorded for the experiments presented in
Figure 3.8. The horizontal bars correspond to the period during which
complete conversion of methane was observed. Shadow area highlights the
additional water evolution with respect to that formed by methane oxidation.
Squares represent first cycles and circles second cycle. Solid line represents the
catalyst temperature (on the right axis).
For the active catalyst, the water production increases with temperature as CH4 conversion
increases. When the temperature is sufficient to completely oxidize the CH4, the water production
is constant. However, with the catalyst deactivated by treatment under wet CH4 oxidation
conditions, a significant excess of water, above that produced by the steady-state oxidation of CH4,
is released from the sample when the temperature is above 700 °C. Although an accurate estimate
62
of this amount is difficult, owing to the steady-state reaction in the background, we monitored an
amount of water from the deactivated catalyst that is slightly higher than the moles of cerium in
the sample. This could suggest that, besides significant formation of cerium hydroxyls on CeO2,
some hydroxyls are also formed on the alumina.
To determine whether other pretreatment conditions could cause the deactivation
observed in Figure 3.7, the CH4 conversion was monitored under dry conditions for longer times
and after high-temperature reduction, with results shown in Figure 3.10. First, the conversion in a
dry mixture of 0.5 % CH4 and 2.0 % O2 at 600 °C showed no decrease from 100 %, even after 14 h.
Neither the high temperatures nor exposure to CH4 and O2, in the absence of H2O, are responsible
for the catastrophic loss in conversion. Next, the catalyst was reduced at 600 °C in 5.0 % H2/Ar.
When the catalyst was again exposed to the dry reaction mixture, 100 % conversion was achieved
after 6 min.
Figure 3.10
Catalytic oxidation of methane. Aging test under dry reaction conditions at
600 °C, followed by reduction in 5.0 % H2/Ar at 600 °C and subsequent purge
in Ar. Conditions: 0.5 % CH4, 2.0 % O2, Ar balance, O2/O2(stoich)=2,
GHSV=200 000 mL g 1 h 1.
63
Flowing Ar over the catalyst at 600 °C for 30 min also had no effect on the subsequent
activity. The effect of catalyst pre-reduction was also examined by means of operando XAFS
experiments demonstrated that Pd is almost completely reduced under 5.0 % H2/N2 already at
150 °C. In addition, the data indicate the presence of 20 % CeIII in the fresh catalyst, in accordance
with the nanometric dimensions of the ceria crystallites (102). This contribution was determined to
be constant during all operando X-ray absorption near edge spectroscopy (XANES) experiments,
either under dry or wet conditions. Upon reduction at 600 °C under 5.0 % H2/N2, the CeIII content
increases up to 45 %. Excellent fitting of the Pd EXAFS data was achieved by considering only the
contribution of Pd fcc phase with a distance of 0.273 nm (0.271 in bulk Pd) in accordance with a
similar study for Pd/PdO on -Al2O3(0.274 nm) (103). In the absence of oxygen (N2 atmosphere),
operando XANES experiments demonstrated that PdO almost fully decomposes to Pd at 600 °C,
even if in close contact with CeO2 (Figure 3.11b). The equilibrium is reversed under dry conditions
leading to the almost immediate reoxidation of 25 % Pd to PdO followed by a further slow increase
during the following reaction time (Figure 3.11c). Only the use of 5.0 % O2/N2 leads to a steep
increase in the Pd to PdO oxidation (Figure 3.11d). Notably the catalytic activity of the pre-reduced
catalyst (Figure 3.10) is fully recovered after a comparable time (7 8 min) indicating that partial
reoxidation of the Pd is sufficient to guarantee high catalytic activities (101).
Figure 3.11
Fraction of PdO obtained from XANES spectra during subsequent
treatments at 600 °C under a) dry reaction conditions (0.5 % CH4, 2.0 % O2,
N2 balance), b) pure N2, c) dry reaction conditions, and d) 5.0 % O2/N2.
64
To understand the changes that occur in the catalyst following CH4 oxidation under wet
conditions, we examined the effect of catalyst pretreatment conditions using FTIR and XANES
measurements, with results from ex situ FTIR shown in Figure 3.12. The spectrum in Figure 3.12a
was obtained on the catalyst that had simply been calcined to 850 °C. After exposure to air and
mild heating in dry He, a rather broad (OH) stretching band is observed in the region of 3700 cm 1.
Aging the catalyst under dry CH4-oxidation conditions (Figure 3.12b), does not lead to significant
modification of the spectrum. However, following treatment of the catalyst in wet CH4-oxidation
conditions (Figure 3.12c), the (OH) hydroxyl band becomes much more intense and centered near
3720 cm 1, a frequency that has been assigned to mono-coordinated and doubly bridged OH
groups bound to ceria. A small shoulder at 3565 cm 1 may be attributable to triply bridging OH
species (104). Exposing the catalyst to dry CH4-oxidation conditions for 1 h (Figure 3.12d) caused
minimal changes; however, heating the sample to 800 °C (Figure 3.12e) greatly reduced the
intensity of the (OH) stretching band.
The fact that this feature is significantly less evident in the case of Pd/Si-Al2O3 (Figure 3.12e-
f) aged under wet conditions suggests that the stable hydroxyls are somehow linked to the
defective - nanostructured ceria. Nevertheless, since the Al2O3 support presents various (OH) band
in the same region (Figure 3.12f), the intense (OH) band in Figure 3.12c and 3.12d cannot be
definitely assigned to cerium hydroxide. The formation of these stable hydroxyls under conditions
where catalyst deactivation occurs and their removal by conditions that reactivate the catalyst
strongly suggest that they are associated with deactivation. The IR spectra of the reference
impregnated Pd-CeO2/Si-Al2O3 samples are different, indicating the influence of the preparation
method on nature of the surface hydroxyls. Remarkably, the wet aging at 600 °C leads to less
marked changes in the population of hydroxyls.
65
Figure 3.12
FTIR spectra of Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 a) fresh sample and after
different treatments: b) dry reaction condition at 600 °C for 4 h, c) wet
reaction condition at 600 °C for 4 h, d) wet reaction condition at 600 °C for 4
h, followed by dry condition at 600 °C for 1 h, e) wet reaction condition at
600 °C for 4 h, followed by dry condition at 800 °C for 1 h, f) Pd/Si-Al2O3 and
conventional impregnated Pd(1 wt%)-CeO2(9 wt%)/Si-Al2O3 after different
treatments g) dry reaction condition at 600 °C for 4 h, h) wet reaction
condition at 600 °C for 4 h.
Results from the operando XANES experiments on Pd(1 wt%)@CeO2(9 wt%)/Si-Al2O3 are
reported in Figure 3.13. Figure 3.13a shows XANES spectra of the Pd edge as a function of time as
the catalyst is exposed to dry CH4oxidation conditions at 600 °C. The spectra remain unchanged
over a period of 79 min; from measurements on reference samples (shown in the inset of Figure
3.13a), it is apparent that the Pd is present as PdO. Upon the introduction of 15.0 % H2O to the
66
reaction mixture (Figure 3.13b), the spectra evolve with time to form a mixture of both PdO and the
less active metallic Pd. As shown in the inset, approximately 60 % of the PdO is converted to Pd
after 80 min under these conditions.
Figure 3.13
Normalized XANES spectra obtained during operando methane oxidation
under a) dry and b) wet reaction conditions at 600 °C. Conditions: 0.5 % CH4,
2.0 % O2, 15.0 % H2O (if present), N2 balance. The insets report XANES of
reference Pd and PdO and PdO percentage during wet reaction conditions
at 600 °C, respectively.
CO chemisorption data on fresh, dry, and wet aged samples, as well as the corresponding
high temperature regenerated ones are reported in Table 3.1. The data clearly indicate that,
consistently with the stable catalytic activity, the relatively high accessible Pd surface area of the
fresh Pd@CeO2/Si-Al2O3sample is not altered by the dry aging. Vice versa, the wet aging at 600 °C
leads to an appreciable decrease of the capability to chemisorb CO. Although significant, the
observed decrease of exposed metal surface area (Table 3.1) cannot alone explain the observed
dramatic deactivation of the catalytic activity. Remarkably, the initial Pd accessibility is fully
recovered after the high temperature treatment, which eliminates the hydroxyls groups and
restores the catalytic activity. The results for the conventional impregnated Pd-CeO2/Si-Al2O3 were
different and the initial accessible Pd surface area is only marginally reduced by either dry or wet
aging (Table 3.1). Only minor sintering occurs with a few hours of aging under the adopted
67
conditions and there is no significant coverage of the catalytic active phase by reduced cerium
oxide or by hydroxyls. However, the initial metal-support contact area, as indirectly indicated by
the metal dispersion, is decreased, partially accounting for the lower activity.
Table 3.1
Accessible Pd surface area, CO/Pd ratio and apparent particle size obtained from
low temperature CO chemisorption on Pd@CeO2/Si-Al2O3 and conventional
impregnated Pd-CeO2/Si-Al2O3.
3.3.3 Discussion
Pd-CeO2 based catalysts show very good low-temperature activity for methane oxidation
(69), especially when prepared in a hierarchical structure that maximizes the contact between the
PdO core the surrounding porous ceria (8). The reactivity of these systems strongly depends on the
oxidation state of Pd, with PdO showing much higher activity than Pd, so that the O2 partial
pressure in the reaction feed plays a major role in affecting rates. The intimate contact between the
Pd phase and CeO2 obtained in our Pd@CeO2/Si-Al2O3 catalysts not only stabilizes Pd particle size
but also facilitates oxygen spillover, reducing the extent of PdO decomposition at high
temperatures and the subsequent decrease in activity associated with the transition of PdO to Pd
(8).
68
At lower temperatures, the presence of water is known to significantly alter the activity of
all Pd catalysts due to formation of hydroxyls on the Pd surfaces (84, 87, 97 99). In conventional
impregnated catalysts, such as Pd/Al2O3, the extent of this deactivation significantly decreases with
increasing temperature, owing to the thermal instability of surface hydroxyls. However, the
presence of water at high temperature can promote Pd sintering (105). Notably, on those catalysts,
the high temperature PdO-Pd phase transition leads also to a significant decrease of catalytic
activity even under dry conditions. What we have observed in the present study is that there is an
additional deactivation process occurring on the Pd@CeO2/Si-Al2O3 core-shell catalysts during
oxidation of methane in the presence of water vapor at high temperatures. After exposure to these
conditions, the exceptional activity observed for Pd@CeO2/Si-Al2O3 was lost. The presence of water
induces 1) a shift of the light off curve to higher temperature and 2) the appearance of the
transient deactivation generally observed in the 600 750 °C range.
The influence of water on catalytic activity can be understood by considering the reaction
mechanism that is generally accepted for palladium supported on CeO2. In the presence of PdO, a
Mars-van Krevelen mechanism is operative, in which PdO lattice oxygen atoms are consumed by
CH4 to produce CO2 and water (106). Re-oxidation of palladium can occur by reaction with O2 from
the gas phase or by spillover of O atoms from CeO2 (85). For each oxygen vacancy formed by
oxygen spillover, two reduced CeIII ions are generated (107) and these in turn must be re-oxidized
by molecular O2. Therefore, the excellent catalytic properties of the core-shell Pd@CeO2 catalyst can
be associated with intimate contact between the PdO phase and the CeO2 nanocrystallites which
leads to efficient O spillover, resulting from the peculiar core-shell structure.
Consistent with previous results (21), the presence of water at medium-low temperature
results in a moderate catalyst deactivation. The effect of water in the reaction media at high
temperature is more remarkable and less reversible (Figure 3.7). IR data shows that, in the presence
of a large amount of water, the oxygen vacancies present on the nanoceria can react with H2O,
forming OH groups on the surface of the ceria nanoparticles (Figure 3.12).
Results from density functional theory (DFT) simulations performed on reduced CeO2 (1 1 1)
indicate that water can dissociate into an OH group, which fills the oxygen vacancy, and an H atom
which bonds to a surface O atom(108, 109). Therefore, the reaction of a vacancy, caused by
reduction of ceria, with one water molecule results in the formation of two hydroxyl groups,
without affecting the oxidation state of CeIII ions close to the original oxygen vacancy site. The
hydroxyl groups formed by the dissociation of water on the oxygen vacancies can in turn inhibit
the oxygen diffusion on the surface of the CeO2nanoparticles and stabilize CeIII (100). Since the
oxygen vacancies originate from the O back spillover from ceria to the metal, a process that
69
involves short range interactions (110), it is reasonable to expect that the hydroxyls resulting from
the reoxidation by water are mostly located in the proximity of the Pd. In this way, an efficient Pd
re-oxidation is hindered and a deactivation of the catalyst occurs (Scheme 3.1), owing to the PdO
transformation to Pd as observed by operando XANES experiments (Figure 3.13).
Scheme 3.1 A representation of the proposed evolution of the Pd@CeO2/Si-Al2O3 catalyst
accounting for the deactivation observed in the presence of water.
The proposed reaction scheme is fully consistent with both the transient and steady state
catalytic data. Under transient conditions (Figure 3.4), the rapid increase in temperature does not
allow a significant buildup of hydroxyl groups, the stability of which decreases with temperature. In
steady-state experiments at 500 °C (Figure 3.6), the weak and fully reversible deactivation is still
consistent with the competitive adsorption of water and methane. At this temperature, the
occurrence of ceria back spillover is not a key requirement for having high activity. The catalytically
active PdO is thermally stable at 500 °C and Pd can be easily oxidized by gas-phase molecular
oxygen. On the other hand, aging at 600 °C under wet reaction conditions (Figure 3.7) induces a
deeper deactivation and a reduction of PdO observed by EXAFS (Figure 3.13), along the formation
of hydroxyl groups (Figure 3.12). The activity recovery observed when switching from wet to dry
reaction conditions at 600 °C (Figure 3.7) is a slow process that involves the recovery of active
phase accessibility and hydroxyl removal, which can be outlined in the following steps:
1. the recombination of OH groups to form a chemisorbed water molecule;
2. the desorption of water leaving an oxygen vacancy on the CeO2 surface;
3. the reaction of O2 from the gas phase with surface O vacancies on CeO x;
4. the migration of O species to Pd forming PdO.
70
Since steps 3 and 4 are also involved in the fast regeneration of the catalyst after reduction
in H2/Ar at 600 °C (Figure 3.10), the slow recovery observed after aging under wet reaction
conditions should be limited by step 1 and/or step 2 (Scheme 3.2). Notably, the same steps are
involved in the reduction of CeO2-based materials in H2 (111, 112) and are recognized as the rate
determining steps for the processes operative above 550 °C (113, 114).
Scheme 3.2 A representation of OH groups recombination to chemisorbed H2O (step 1)
and of water desorption, with an oxygen vacancy formation (step 2).
Investigations of the catalyst that was deactivated under wet conditions at 600 °C (Figure
3.8) further corroborate the proposed mechanism of high temperature deactivation. Under dry
conditions, the light-off curves for the aged sample are shifted significantly towards higher
temperatures (Figure 3.8). In addition to the water generated by methane oxidation, the additional
water evolved from the deactivated catalysts in the temperature range from 700 to 850 °C (Figure
3.9) confirms the high temperature stability of the hydroxyls groups on CeIII and their role in the
deactivation phenomenon. The catalyst deactivated under wet condition is, therefore, composed
of a mixture of PdO and metallic palladium (EXAFS data of Figure 3.13) surrounded by a shell of
hydroxylated (IR data of Figure 3.12) and partially reduced ceria. In addition to the specific lower
activity of Pd with respect to PdO, the presence of thermally stable OH may also hinder gaseous
reactants from reaching the surface and limit the active sites accessibility, as observed by the
decrease of CO chemisorption capability (Table 3.1). Finally, the exceptional high activity observed
under dry condition on the fresh sample is fully recovered on the wet aged sample after complete
removal of the hydroxyls on CeO2 (Figure 3.8).
3.3.4 Strategies to Improve Catalytic Stability and Conclusions
The amount of Ce in the Pd@CeO2 synthesis can be varied by changing the ratio in which
the two pre-formed building blocks are mixed (functionalized Pd NPs and Ce alkoxide). A Pd (1
71
wt%)@CeO2(6 wt%)/Si-Al2O3 catalyst was prepared and tested in methane oxidation under wet
conditions at 600°C (Figure 3.14). The catalyst was not only very active, but also remarkably more
stable than the Pd@CeO2(1:9 wt), retaining 70% of conversion after 16 hours of wet conditions
operation. This was due to a much higher active phase accessibility after wet aging ( 2.7 m2 g-1
versus 1.1 m2 g-1 for the wet aged 1:9 catalyst). For comparison, impregnated catalysts having
comparable formulation, i.e. Pd(1 wt%)-CeO2(6 wt%)/Si-Al2O3 and Pd(1 wt%)-CeO2(9 wt%)/Si-Al2O3
were tested under the same conditions. As expected, the Pd(1 wt%)-CeO2(6 wt%)/Si-Al2O3 was less
active than the Pd(1 wt%)-CeO2(9 wt%)/Si-Al2O3 catalyst, since the contact between the promoter
and the active phase is statistically less probable. Both the impregnated catalysts were quite stable
under the reaction conditions, in accordance with the chemisorption results, indicating that the
active phase is still accessible even after wet aging (Table 3.1). These results show that hierarchical
catalysts are deactivated by water in a different way with respect to impregnated catalysts, and
that the strategies to enhance the catalytic performance of nanostructured catalysts may be
different.
Figure 3.14
Steady state aging test under wet reaction conditions at 600 °C for Pd(1
wt%)@CeO2(9 wt%)/Si-Al2O3 (solid squares); Pd(1 wt%)@CeO2(6 wt%)/Si-
Al2O3 (empty squares); Pd(1 wt%)-CeO2(9 wt%)/Si-Al2O3 (solid line); Pd(1
wt%)-CeO2(6 wt%)/Si-Al2O3 (dashed line). Conditions: 0.5 % CH4, 2.0 % O2,
15.0 % H2O, Ar balance, O2/O2(stoich)=2, GHSV=200 000 mL g 1 h 1.
72
In conclusion, methane oxidation on Pd@CeO2/Si-Al2O3 is strongly influenced by the
presence of water during methane oxidation at higher temperatures. Although the catalyst is
thermally stable under both run-up experiments and steady-state experiments at both 500 and
600 °C under dry conditions, the addition of water progressively deactivates the system. While at
temperatures below 500 °C deactivation is easily reversed by removing the water, at higher
temperatures an irreversible deactivation process is observed. This is attributable to lower active
phase accessibility and to the formation of stable hydroxides, which are decomposed only by high-
temperature treatment. Lowering the amount of CeO2 in hierarchical catalysts leads to enhanced
stability thanks to the retention of more accessible active phase. The results show that design of
catalysts for methane oxidation must maximize metal-support interactions to favor oxygen transfer
from a reducible promoter, while keeping the active phase accessible to gas phase reactants. It has
also been demonstrated that hierarchical Pd@ZrO2/Si-Al2O3catalyst can both guarantee high
activity thanks to the nanostructural design and also good hydrothermal stability (115).
73
3.4 Phosphorus Poisoning
3.4.1 Introduction
Phosphorus is a particularly serious poison for CeO2-promoted catalysts and is a primary
agent for irreversible deactivation of automotive catalytic converters under real operating
conditions (116 118). The presence of phosphorus compounds (P2O5 or H3PO4) in vehicle exhaust is
due to decomposition/volatilization of motor oil anti-wear additives, such as zinc
dialkyldithiophosphate (ZDDP) (35, 119). These additives are present in most available motor oils in
concentration up to 1-2% (although this can vary depending on the final application) and their
effect on catalytic converters is well documented (116 118). Based on studies of model systems
(CeO2 and CexZr1-xO2) (34 37, 119 122), it is known that mixed phosphates (Zn, Ca and Mg) form
glassy overlayers on the washcoat surface and also react with the catalyst to form CePO4 and AlPO4
(122).
CePO4 formation is irreversible and detrimental to the catalytic activity due to loss of
oxygen storage capacity (OSC), which is caused by locking of the Ce3+/Ce4+ pair in the 3+ state (121,
122). Both the direct reaction of CeO2 with phosphorus compounds in the gas phase and the
reaction of CeO2 with P-containing species on the catalyst (e.g. aluminum phosphate) can lead to
the formation of CePO4 (37). Based on evidence from total reflection X-ray florescence (TXRF), XPS
and 31P NMR data on P-poisoned CeO2, phosphates species formed on the surface and sub-surface
region dramatically decrease OSC. Cerium phosphate is also very stable and cannot be removed
from either pure CeO2 or CexZr1-xO2 by calcination treatments to 1000 °C (34, 123). A few studies
indicated partial removal of phosphate species by washing the spent catalyst with oxalic acid (119)
or chlorine containing species (124). However, CePO4 persists even after these treatments (119),
implying that alternative methods need to be developed to address the problem of P-poisoning.
Despite previous work on regenerating P-poisoned catalysts, relatively little is known about
the conditions in which CePO4 is formed. In studies regarding commercial catalysts, the effect of P-
aging is evaluated by comparing fresh samples and samples aged for 30, 000 - 160,000 km in
conventional automobiles (116 118). In model systems studies the aging treatment is typically
simulated by depositing phosphates by impregnation of the catalyst with NH4H2PO4 solutions
followed by calcination to get a final P content of 0.04-4.5 wt% (34 36). In one study, CeO2 aging
was performed for only 10 hours by introducing 85 ppm H3PO4 through the gas feed (37). Notably,
aging effects were similar in model systems and real catalysts, even if the chosen conditions, the
materials studied and method of P addition were different. This suggests that model systems
studies are relevant to real applications, and also that aging can be very fast as soon as phosphates
reach the catalytic bed at a certain temperature of formation. Despite this, to our best knowledge,
74
time-on-stream deactivation studies on commercial or model systems are still too limited. Also, the
temperature threshold at which CePO4 is formed was only reported by Xu et al. (37) for pure CeO2
under lean conditions (600°C). However, in their study, gas-phase P2O5 was introduced by thermal
decomposition of aqueous H3PO4, resulting in 5% H2O in the reaction atmosphere. In this section,
we show that the introduction of H2O in methane oxidation reaction mixture enhances the effect of
phosphorus poisoning, leading to deactivation. This is particularly relevant to lean-burn engines
converters and other real applications, in which water is typically present in 5-15% vol.
concentration (89, 91).
3.4.2 Results
In this study, model catalysts based on Pd@CeO2 units deposited on graphite foils were
prepared, in order to have conductive supports suitable for PES analysis. A Pd:CeO2 ratio of 1:6 in
weight was chosen, accordingly to previous results (see Figure 3.14, chapter above). Two types of
graphite foils were used: one pure (99.8% purity, phosphorus-free graphite), and one containing
phosphorus (99% purity). The foils were cut into 9x9 mm slides and analyzed by EDS to determine
their P content. As expected, no P was observed on the pure graphite support, while the P-graphite
samples contained 1000 ± 100 ppm of P. The P signal did not change appreciably upon deposition
of Pd@CeO2 units, nor was it affected by any of the aging treatments performed in this study.
Complementary XPS/SRPES analysis was performed on the as-received graphitic supports
to determine the chemical state of phosphorus and the surface/bulk distribution. Prior to
calcination, there was no P signal on either the P-graphite or the pure graphite. A phosphorus
signal, at a Binding Energy (BE) corresponding to that of PO43- (133 eV) was only observed on the P-
graphite after thermal treatment to 450 °C in air (see Figure 3.15). Similar treatment of the pure
graphite did not cause the appearance of a P signal. This suggests that P is initially present only in
the bulk of the P-graphite support and is released during the thermal treatment. An O 1s signal at
530.5 eV appeared together with the P signal, indicating the presence of PO43- species. This
fingerprint of PO43- was present on all Pd@CeO2/P-graphite samples, independent of aging
conditions and with a similar intensity as was observed following calcination at 450 °C. In
accordance to these observations, phosphates can be present as intercalating compounds of
graphite foils manufactured by squeezing exfoliated graphite (125).
75
Figure 3.15 P 2p XPS spectra of P-graphite samples: fresh support (A), 450°C calcined
support (B), Pd@CeO2/P-graphite samples fresh and aged (C).
SEM images of the fresh Pd@CeO2/graphite and Pd@CeO2/P-graphite catalysts are shown in
Figure 3.16 and exhibit very similar surfaces, displaying large, smooth domains of a continuous
film, separated by small, shallow cracks. The distributions of Pd, Ce and O signals over the samples
were homogeneous, demonstrating that the Pd@CeO2 particles were well-dispersed and exposed
to the gas phase for both supports, an important prerequisite for model catalytic systems.
Figure 3.16
20 m viewfield SEM images and EDS mapping of Pd, Ce and O signals of
fresh Pd@CeO2/graphite (top row) and Pd@CeO2/P-graphite (bottom row)
samples.
76
conditions, at 500 °C and 600 °C, in dry and wet conditions. In all experiments, the catalyst was
ramped to the desired temperature in 0.5% CH4, 2.0% O2 and 15.0% H2O (for wet conditions) and
held at that temperature for the chosen time span. After aging, the gas flow was switched to pure
Ar and the samples cooled to room temperature before being transferred to either the SRPES/XPS
line or to the SEM.
As shown in Figure 3.17A, the fresh model Pd@CeO2/graphite catalyst showed significant
and constant evolution of CO2 as result of CH4 oxidation at both 500 and 600 °C. While rates were
stable under dry conditions, they decreased with time at both temperatures when water was
added. The Pd@CeO2/P-graphite samples were less active for methane oxidation (Figure 3.17B).
Most notably, there was a fast and irreversible deactivation during wet aging at 600 °C. After 1 h of
isothermal treatment under these conditions, the sample was almost completely inactive.
Figure 3.17
CO2 evolution over time. A: Pd@CeO2/graphite; B: Pd@CeO2/P-graphite.
Circles: 600 °C aging; squares: 500 °C aging; filled symbols: dry conditions
(CH4 0.5%, O2 2%, GHSV: 106 mL g-1 h-1) ; open symbols: wet conditions (15%
H2O vapor added to reaction atmosphere).
77
SEM images of the aged samples in Figure 3.18 show differences between the catalysts
supported on graphite and P-containing graphite. The surfaces of the Pd@CeO2/graphite samples
were not appreciably altered by any of the aging treatments. However, images of the Pd@CeO2/P-
graphite samples show features having diameters of 10-30 nm, with sizes that are slightly bigger
for 600 °C aging treatments. These spots are partially sintered Pd@CeO2 particles. The SEM analysis
suggests that the presence of phosphorus causes partial aggregation of the Pd@CeO2 units and
growth of crystallite sizes that does not occur in the absence of P.
Figure 3.18
SEM of Pd@CeO2/graphite (A-E) and Pd@CeO2/P-graphite samples (A'-E').
The samples were fresh (A, A') or treated for 9 h under the following
conditions: dry reaction conditions at 500 °C (B, B') or 600 °C (D, D'); wet
reaction conditions at 500 °C (C, C') and 600 °C (E, E').
78
Figure 3.19 indicates that the Ce 3d region of the XPS spectra is very different for the
Pd@CeO2/graphite and Pd@CeO2/P-graphite samples. Spectra on the Pd@CeO2/graphite samples
were all similar, with features typical of pure CeO2. Initially, the apparent O:Ce stoichiometry is 1.94
(12% Ce3+) but it increases to 2.0 with aging time. Apparently, a small fraction of the cerium is in the
Ce3+ state in the initial particles but all of the Ce atoms are converted to Ce4+ under the aging
conditions used in this study. This behavior differs from that observed for Pd@CeO2/Si-Al2O3
catalysts, which showed a low but constant Ce3+ fraction for all the treatments (89). Various factors
could contribute to this, including the different surface sensitivity of the techniques used (XPS-XAS)
and the different Pd:Ce ratio, but the different calcination temperatures used here (450 °C vs. 850
°C) and the different supports (Al2O3 vs. graphite) are likely the primary causes.
Figure 3.19
Representative Ce 3d XPS spectra of: (A) the fresh Pd@CeO2/graphite
sample; (B) fresh Pd@CeO2/P-graphite; (C) Pd@CeO2/P-graphite aged at 500
°C; (D) Pd@CeO2/P-graphite aged at 600 °C. Dry conditions are shown by red
lines and wet conditions by the blue lines. The areas of the presented
spectra have been normalized after subtraction of Shirley background and
the curves have been offset for clarity. The positions of the most prominent
peaks in Ce4+ and Ce3+ spectra are marked by grey lines.
79
On the Pd@CeO2/P-graphite sample, in contrast, the XPS spectra show mainly features of
CePO4 (35, 126), even on the fresh sample, for which Ce4+ is also still observed. Spectra taken after
heating to 500 °C under dry and wet aging were similar to each other and showed only small
changes compared to the fresh sample. On the other hand, aging treatments at 600 °C resulted in
dramatic changes. After 30 min of wet aging at this temperature, the Ce4+ signal completely
disappeared. Spectra obtained after prolonged dry aging (9 h) showed partial oxidation to Ce4+
with respect to the fresh sample.
Figure 3.20 shows the percentage of Ce3+ on the Pd@CeO2/P-graphite sample for the
different aging conditions. Since, in the absence of P, the Ce in the Pd@CeO2 particles is almost
completely oxidized to Ce4+ after the initial calcination and after all aging treatments, the trends
observed for the Ce3+ percentage in this case are indicative of CePO4 formation on the surface of
the particles sampled by XPS. The fresh sample contains 75% Ce3+, demonstrating that the particles
are extensively covered by CePO4 already after calcination. For aging at 500 °C, the percentage of
Ce3+ does not change appreciably, regardless of the aging conditions (dry or wet). However, aging
at 600 °C changed the sample dramatically. Wet aging caused an increase in Ce3+ percentage to
100% after 30 min, while dry aging decreased the Ce3+ content, to 40% Ce3+ after 9 h. These results
indicate a crucial role for water in the formation, accumulation, and stabilization of phosphates on
ceria.
Figure 3.20
Calculated Ce3+ percentage for Pd@CeO2/P-graphite samples, determined
by fitting of Ce XPS spectra. Fresh samples: star. Squares: aged at 500 °C
(filled: dry conditions, empty: wet conditions). Circles: aged at 600 °C (filled:
dry conditions, empty: wet conditions). The solid lines are guides to the eye.
80
To gain further insights into the extent of CePO4 formation and the extent of cerium
reduction, operando and ex-situ XANES measurements were performed on the Pd@CeO2/P-graphite
catalysts, with representative results shown in Figure 3.21. In all cases, only minor differences were
observed between fresh, dry-aged and wet-aged samples, at both 500 °C and 600 °C. In contrast to
the XPS results, the Ce3+ percentage did not evolve during aging treatments and ranged from 15%
to 18% in all the samples studied. Indeed, these results are similar to recent in-situ EXAFS data on a
Pd@CeO2/Si-Al2O3 powder having a similar Pd@CeO2 composition with no P poisoning. That study
also indicated the presence of 20% Ce3+ in the fresh catalyst (89).
Figure 3.21
Operando XANES spectra of Pd@CeO2/P-graphite sample at Ce LIII edge: (A)
fresh; (B) during dry aging at 600 °C; (C) during wet aging at 600 °C.
Reference spectra: CeO2 (D), CePO4 (E).
81
The dramatic difference in the valence ratios calculated from XPS and XANES implies that
the surface and bulk concentrations can be very different on these samples. Since the inelastic
mean free path in CeO2 of the photoelectrons originating from Ce 3d level in the XPS study is
approximately 1.2 nm and the expected dimensions of the CeO2 crystallites in the Pd@CeO2
particles is 3 to 4 nm (8), the large difference between surface and bulk concentrations cannot be
explained by the presence of CePO4 at the surface of the initial Pd@CeO2 core-shell particles.
Rather, the data indicate that thermal sintering or partial aggregation of ceria crystallites must
occur under wet aging at 600 °C.
To further investigate the evolution of the Pd@CeO2 particles dimensions during aging, the
Pd@CeO2/graphite and Pd@CeO2/P-graphite samples were characterized by AFM topographic
analysis. Figure 3.22 shows images and height profiles for the Pd@CeO2/P-graphite sample, both
fresh and after 9-h aging at 600 °C. Results for the Pd@CeO2/graphite sample are not shown but
were essentially identical to that observed for fresh Pd@CeO2/P-graphite, even with various aging
treatments. For both Pd@CeO2/graphite and Pd@CeO2/P-graphite, the fresh and dry aged samples
showed small features, approximately 10 nm in height, similar to what was observed for Pd@CeO2
particles deposited onto YSZ(100) single crystals (95) (Figure 3.22A and B). Wet aging did not
appreciably change the Pd@CeO2/graphite sample at either 500 °C or 600 °C. However, wet aging
at 600 °C caused dramatic sintering on the Pd@CeO2/P-graphite sample, leading to the appearance
of features that were 50-60 nm high and 50-70 nm wide, distributed over a corrugated surface
(Figure 3.22C). The changes in particle size help explain how the bulk and surface compositions can
be so different.
82
Figure 3.22
AFM topography images with representative line scans for Pd@CeO2/P-
graphite: (A) fresh; (B) aged at 600 °C under dry conditions for 9 h; (C) aged
under wet conditions at 600 °C for 9 h. Please note the different scales in
part C.
Figure 3.23 shows XPS/SRPES spectra for the O 1s region of the Pd@CeO2/graphite and
Pd@CeO2/P-graphite samples and the results support the conclusions reached from the Ce 3d core
level. The spectrum of the fresh Pd@CeO2/graphite sample, Figure 3.23A, shows two peaks at 529.2
eV and 531.6 eV. The peak at lower BE can be assigned to bulk CeO2, while higher BE peak is likely
due to hydroxyl species (127). However, the presence of other species having similar BE cannot be
entirely ruled out. Carbonates arising from reaction with the graphite supports are not expected to
form during the thermal treatments performed here; also, similar O 1s spectra were observed for
Pd@CeO2 on Au (not reported here). When the Pd@CeO2/graphite sample was aged under wet
conditions, Figure 3.23B-C, the intensity of the peak at 532 eV was enhanced and a slight shift to
higher binding energies was observed, further supporting the assignment of the signal to hydroxyl
species (85, 89, 127). With Pd@CeO2/P-graphite, the spectrum of the fresh sample, Figure 3.23B',
again showed a peak near 529 eV due to bulk CeO2 but the largest peak was centered at 530.4 eV,
83
which can be assigned to CePO4 basing on the preliminary analysis performed on the P-containing
graphite. There was very little change in the spectrum of samples aged under dry conditions at
either 500 °C or 600 °C, but wet aging at either temperature gave increased intensity in the region
assigned to hydroxyls. After wet aging at 600 °C, the peak associated with bulk CeO2 essentially
disappeared. This is consistent with the disappearance of the Ce4+ signal in the Ce 3d region of the
600 °C wet-aged samples.
Figure 3.23
Left: representative O 1s XPS spectra of Pd@CeO2/graphite samples: fresh
(A), 500°C aged 9h (B), 600°C aged 9h (C); Right: representative O 1s XPS
spectra of : (A') fresh Pd@CeO2/graphite; (B') fresh Pd@CeO2/P-graphite; (C')
Pd@CeO2/P-graphite aged at 500 °C; (D') Pd@CeO2/P-graphite aged at 600
°C. The red lines were obtained after aging under dry conditions and the
blue lines under wet conditions. The areas of the presented spectra have
been normalized to the CePO4 contribution (B,C and D) and the P free
samples have been adjusted to comparable intensity. The curves have been
offset for clarity. The positions of assigned O 1s contributions are marked
with grey lines.
84
XPS spectra of Pd 3d region on representative samples are shown in Figure 3.24. For both
Pd@CeO2/graphite and Pd@CeO2/P-graphite, the fresh samples show two peaks at 337.2 eV and
342.4 eV, which are almost certainly due to PdO. The two peaks are due to the Pd spin-orbit split
doublet (Pd 3d5/2 and Pd 3d3/2). The BE of the doublet is actually close to the BE expected for PdO2
(128); however, a shift to higher binding energy is often observed for metal and metal-oxide
nanoparticles, including Pd and PdO (129 135). For example, the 3d5/2 signals of both PdO and Pd
nanoparticles have previously been reported to shift to approximately 1.0 eV higher BE due final-
state effects (129, 133).
Figure 3.24
Representative Pd 3d XPS spectra: (A) Pd@CeO2/graphite; (B) fresh
Pd@CeO2/P-graphite; (C) Pd@CeO2/P-graphite aged at 500 °C; (D)
Pd@CeO2/P-graphite aged at 600 °C. Aging under dry conditions is indicated
by the red lines while wet aging is indicated in blue. The curves have been
offset for clarity.
85
The fact that our samples contained PdO was confirmed by operando and ex-situ XANES
measurements on both the graphite and P-graphite supported catalysts (Figure 3.25). Neither wet
nor dry aging had any effect on the XPS spectrum for Pd on the Pd@CeO2/graphite sample;
however, the same was not true for Pd@CeO2/P-graphite. While dry aging did not affect the
spectrum, wet aging of the P-containing sample at 600 °C resulted in the complete disappearance
of the Pd signal. Since EDS and XANES results demonstrate that Pd is not lost under these
conditions (Figure 3.25), the loss of Pd signal in XPS must be due to burial of Pd under the CePO4.
This agrees with the SEM results, showing that the Pd@CeO2 particles undergo severe sintering
during wet aging when P is present.
Figure 3.25
Pd K edge XAS spectra of: (A) fresh (black line), 600°C dry aged 9h and 600°C
wet aged 9h Pd@CeO2/P-graphite samples, not normalized; (B) normalized
PdO reference.
3.4.3 Discussion
One of the main objectives of this work was to study the effect of P poisoning on the
catalytic performance of Pd-CeO2 catalysts. Similarly to other P-poisoning studies, phosphorus was
deliberately introduced in the system during the catalyst preparation (34 36), rather than
introducing it from the gas feed during aging treatments (37). In this way, H2O is not introduced in
the reaction mixture by decomposition of H3PO4 to P2O5 and the effect of water addition can be
studied separately. The combined results in this study demonstrate that the presence of
phosphorus in the vicinity of a Pd/ceria catalyst results in the rapid formation of CePO4 at
86
temperature as low as 450 °C. Since vapor pressure of phosphates is negligible under these
conditions, the formation of CePO4 indicates that there is a high affinity of Ce and P. This agrees
with previous results from Xu et al., who observed formation of CePO4 from the reaction of CeO2
with AlPO4 (37). Notably, the distribution of CePO4 in the surface and subsurface of CeO2 particles is
in agreement with other observations from the literature (35 37, 121, 122), even if the source of
phosphorus and aging conditions were different.
Interestingly, the phosphorous-poisoned catalysts maintain a relatively high activity under
dry conditions, despite being extensively covered by phosphates. Indeed, the XPS spectra even
indicate that cerium phosphate can be partially removed from the surface of ceria particles at 600
°C under dry conditions. In agreement with this, López Granados et al. reported that, on P-
poisoning of CeO2 by addition of (NH4)2HPO4, followed by calcination to 600 °C (35) some of the
CeO2 surface was not converted to CePO4 and was still available for oxygen exchange with gas-
phase oxygen, even for samples having a high P:Ce ratio.
The presence of water changes things completely, and a rapid, irreversible deactivation is
observed at temperatures above 600 °C. At these temperatures, CePO4 becomes highly mobile and
causes severe aggregation of ceria particles and Pd encapsulation. The results of this study indicate
that water and phosphorus have a cooperative effect and take part in a deactivation mechanism
activated by high temperature. For the first time, the deactivation of Pd-CeO2 catalysts in the
presence of P was studied in steady-state experiments, revealing that poisoning occurs in very
short time under conditions relevant for real applications (36, 37, 122). Although surface blockage
by CePO4 is reported in the literature as the main effect of P-poisoning on ceria-based catalysts, our
work reveals that phosphorus can also cause deep morphological transformations of the catalyst
and dramatic loss of activity, especially in the presence of water.
3.4.4 Conclusions
In conclusion, the effect of phosphorus poisoning on the catalytic oxidation of methane
over Pd@CeO2/graphite catalysts was found to be dramatically influenced by temperature and
presence of H2O. When P was not introduced in the catalyst formulation, the catalysts were active
and stable under all studied conditions (500-600 °C; dry and wet conditions). On the other hand, P-
poisoned catalysts were less active and stable because of partial thermal sintering. Water vapor
causes rapid and complete deactivation at higher temperatures by inducing severe aggregation of
ceria nanoparticles, incorporation of Pd active phase in the bulk of the crystallites and exposure of
CePO4 to the catalyst surface. The combination of XPS/SRPES, operando XANES measurements,
SEM/EDS and AFM techniques provide evidence of a temperature dependent, water-driven P-
poisoning of Pd and CeO2-based oxidation catalysts.
87
3.5 SO2 Poisoning
3.5.1 Introduction
SO2 is a particularly serious poison of catalytic converters, produced by the oxidation of
sulfur compounds present in fuels and in lubricating additives (23, 136). Despite the continuing
reduction of sulfur in fuels (The Ultra-Low Sulfur Diesel (ULSD) regulation established a 15 ppm
sulfur content for diesel fuel in 2006(137)), long-term exposure to low concentrations of SO2 is still
detrimental to the catalytic activity of the exhaust after-treatment catalysts (21, 136, 138). Under
lean conditions and at temperatures above 200 °C, the presence of SO2 in the exhaust leads to
formation of sulfate species on both the support and the active phase (71, 139). One of the possible
pathways for the formation of sulfates involves SO2 oxidation to SO3 by O2 (Reaction 3.1) and
subsequent adsorption of SO3 on the metal-oxide surface (Reaction 3.2).
(3.1)
(3.2)
Alternatively, SO2 may be oxidized by oxygen from the support or disproportionate to SO
and SO3 (140). Formation of sulfates on supports such as alumina can also deactivate the catalyst by
acting as a buffer of SOx that prevents adsorption of SO2 onto the active phase during exposure to
gas-phase SO2 but prolongs the poisoning effect after SO2 removal due to a slow decomposition of
the accumulated sulfates(23, 136). With a non-sulfating support, such as SiO2, the active phase will
not be protected from poisoning, leading to faster deactivation, but also to faster recovery (23).
SO2 poisoning on reducible supports, CeO2 and CexZr1-xO2, has been widely investigated
because of the importance of these oxides in catalytic converters (141 144). It is known that ceria
forms sulfates that are stable to relatively high temperatures (600 °C - 700 °C) (142, 145). ZrO2 forms
fewer sulfates when exposed to the same conditions, and these are mostly confined to its surface
(142). However, conflicting observations have been made regarding the resistance of Ce-Zr (CZ)
mixed oxides to SO2 in comparison to CeO2 and ZrO2. Luo et al. reported that CeO2 and CZ catalysts
were affected in a similar manner by SO2 poisoning for both the water-gas-shift (WGS) and CO-
oxidation reactions (142), while Deshmukh et al. reported enhanced resistance to poisoning for the
mixed oxides (144).
In this chapter, we report the effect of SO2 poisoning on methane oxidation over
hierarchical Pd@CexZr1-xO2 catalysts supported on modified alumina. The previously reported
synthesis (92) was modified in order to achieve a range of shell compositions, from pure CeO2 to
88
pure ZrO2. We investigate the effect of SO2 on these core-shell catalysts using XPS on model
catalysts designed to closely resemble the high-surface-area (HSA) materials in both surface
composition and temperature treatments, while avoiding charging and low resolution problems
often experienced with powdered materials (146).
3.5.2 High-Surface-Area Materials
The synthesis of Pd@CeO2 and Pd@ZrO2 particles reported in previous works (8, 92) was
modified in this study to allow the preparation of Pd@CexZr1-xO2 particles. Ce and Zr alkoxides were
mixed together before slowly adding the dispersed Pd-MUA nanoparticles to allow the reaction of
both alkoxides with the carboxylic group of MUA. The Pd:(Ce+Zr) molar ratio was kept constant
and the following Ce:Zr molar ratios were selected to investigate the generality of the synthetic
method: (80:20), (60:40), (40:60), (20:80). Pure Pd@CeO2 and Pd@ZrO2 units were also prepared as
references. After controlled hydrolysis, the particles were deposited on hydrophobic, silanized
alumina and the materials were calcined at 850 °C for 5 h. The final Pd@MOx/Si-Al2O3 catalysts were
characterized by XRD to check for the formation of mixed or segregated CZ oxide phases (Figure
3.26 and Figure 3.27).
Figure 3.26
XRD patterns of the Si-Al2O3 calcined at 850 °C for 5 h (grey line) and
Pd@MOx/Si-Al2O3 samples calcined at 850 °C for 5 h (orange: CeO2; green: CZ
80:20; red: CZ 60:40; magenta: CZ 40:60; purple: CZ 20:80; blue: ZrO2).
89
Figure 3.27
XRD patterns of hydrophobic the Si-Al2O3 calcined at 850 °C for 5h (a) and
Pd@MOx/Si-Al2O3 samples calcined at 850 °C for 5h: ZrO2 (b), CZ 20:80 (c), CZ
40:60 (d), CZ 60:40 (e), CZ 80:20 (f), CeO2 (g).
A complete Rietveld analysis of the CZ XRD pattern was not possible due to overlap with
the signal from the Al2O3 support, which was more than 90 wt% of the studied materials and
- -Al2O3 after thermal pre-treatment (147). Nonetheless,
the partial XRD spectrum reported in Figure 3.26 clearly showed the first and most intense
reflection of the CZ mixed oxides, which varied between 28.5° (for CeO2) to 30.2° (for ZrO2).
Therefore, only pseudo-cubic cell parameters were calculated in the present case, despite the fact
90
that a transition from cubic to tetragonal crystal structure is expected in the oxide promoter with
high ZrO2 content (148). The linear dependence of the cell parameter for the pseudo-cubic cells
with the CZ composition is consistent with the formation of solid solutions between CeO2 and ZrO2
(inset of Figure 3.26) (149). Only in the case of CZ 40:60 did a shoulder appear at lower diffraction
angles, suggesting that segregation of a small amount of a CeO2-rich phase occurred for that
sample. Phase separation has been observed previously for CZ mixed oxides with 40:60 Ce:Zr
ratios, because this composition is thermodynamically unstable(148). The decreased intensity of
the main reflection with increasing ZrO2 content is due to the lower weight loading and scattering
factor of Zr.
The mean crystallite sizes of the CZ mixed oxides were calculated to be in the range of 7-11
nm (Table 3.2), which is in good agreement with TEM results (5-10 nm apparent particles size).
Notably, the apparent size of CZ mixed-oxide crystallites is smaller than that of CeO2, as expected in
view of the stabilizing effect of Zr(149). The most intense reflection for PdO, the (101) plane, can be
observed at 34°. The fitting of this line is complicated by the overlapping of other reflections, so
that the mean crystallite size calculated by the Scherrer equation (15-18 nm, see Table 3.2) is not
very indicative of the actual size distribution of the PdO active phase in the samples, as discussed
further in the section on TEM analysis.
Table 3.2 Cell parameter and apparent crystal size of CZ mixed oxides and apparent
crystal size of PdO.
91
Physisorption and chemisorption experiments on Pd@MOx/Si-Al2O3 catalysts revealed that
all the samples had similar surface areas, pore-size distributions, and Pd accessibility, as detailed in
Table 3.3.
Table 3.3 Results of N2 physisorption and CO chemisorption analysis of the
investigated samples.
The results are comparable with those reported in previous studies on Pd@CeOx/Si-Al2O3
and Pd@ZrOx/Si-Al2O3 (89, 115). On the other hand, the Pd dispersion for the Pd/Si-Al2O3 was
significantly lower due to severe sintering of Pd nanoparticles. The pore-size distributions were also
consistent with previous results: Pd/Si-Al2O3 and the Si-Al2O3 support have similar pore structures,
with pore sizes in the range from 10 to 50 nm in diameter, while the Pd@MOx/Si-Al2O3 catalysts
exhibited fewer large pores due to partial filling with the nanostructured units, along with the
formation of small pores (around 10 nm in diameter) associated with the units themselves (Figure
3.28).
92
Figure 3.28
BJH pore size distribution taken from the desorption branch of: Si-Al2O3
(grey), Pd/Si-Al2O3 (black) and Pd@MOx/Si-Al2O3 (orange: CeO2; red: CZ
60:40; blue: ZrO2).
Representative TEM-EDS results (Figure 3.29) demonstrated that all Pd@MOx-based
catalysts have similar morphologies: the CZ particles, identified by EDS spectra and lattice fringes,
are 5 10 nm in diameter and are sometimes aggregated. For Pd@CZ 60:40, the signals of Ce and Zr
observed in EDS mapping mode are always associated and the signal analysis in spot mode
revealed a good agreement with the desired Ce:Zr stoichiometric ratio. The Pd signal in EDS is
typically low and diffuse, though some large palladium particles were observed, similar to what
was reported by Zhang et al. for similar catalysts (150). Such a bimodal particle distribution is not
detected by XRD, since the very small PdO nanoparticles will give a broad reflection that is hard to
distinguish in the presence of overlapping XRD patterns from the Al2O3 support. The Pd/Si-Al2O3
catalyst showed only very large Pd particles (up to 100 nm in diameter) due to severe sintering
after calcination at 850 °C for 5 h. In all cases, no apparent changes in morphology were observed
after any SO2 aging treatments, discussed in the following section (Figure 3.30).
93
Figure 3.29
Representative High-Angle Annular Dark-Field (HAADF) images (left
column) and EDS mapping (right column) of samples Pd@CeO2/Si-Al2O3
(Pd@CeO2), [email protected]/Si-Al2O3 (Pd@CZ), Pd@ZrO2/Si-Al2O3 (Pd@ZrO2)
and Pd/Si-Al2O3. EDS mapping colors: blue = Al, red = Pd, yellow = Ce, green
= Zr.
94
Figure 3.30
Representative High-Angle Annular Dark-Field (HAADF) images of samples:
Pd@CeO2/Si-Al2O3 (Pd@CeO2), [email protected]/Si-Al2O3 (Pd@CZ), Pd@ZrO2/Si-
Al2O3 (Pd@ZrO2) and Pd/Si-Al2O3. Left column: samples calcined at 850 °C;
right column: samples aged at 600 °C for 2h in SO2 + dry conditions (2% O2,
0.5% CH4, 50 ppm SO2, balance Ar).
95
Methane-oxidation, light-off experiments performed on each of the Pd@MOx-based
catalysts showed comparable results that were similar to those previously reported for the single-
oxide catalysts (Figure 3.31) (8, 92). By comparison, the Pd/Si-Al2O3 catalyst showed a higher light
off temperature and the usual conversion hysteresis after heating to high temperature (Figure
3.31).
Figure 3.31
Methane oxidation light-off curves over Pd@MOx/Si-Al2O3 (red curve) and
Pd/Si-Al2O3 (black curve). Conditions: 0.5% CH4, 2.0% O2, Ar balance,
GHSV=200000 mLg-1h-1, heating and cooling rates 10 °C min-1.
While light-off curves were comparable for all Pd@MOx catalysts, sulfur tolerance was
expected to vary more strongly with composition. In order to study SO2-poisoning resistance,
aging tests were performed in which the catalysts were exposed to specified concentrations of SO2
(typically 50 ppm) at different temperatures (from 300 to 600 °C). Introducing 50 ppm of SO2 into
the feed under dry conditions (0.5% CH4; 2% O2; Ar balance) caused complete and irreversible
deactivation of all the studied catalysts between 300 °C and 400 °C (Figure 3.32). At 450 °C, each of
the hierarchical, core-shell catalysts was partially regenerated under dry conditions, recovering 50
to 60 % of their initial conversion. The final conversion trend was Pd@ZrO2/Al2O3 >
[email protected]/Al2O3 > Pd@CeO2/Al2O3. Although less active, the Pd/Si-Al2O3 catalyst was almost
completely regenerated under dry conditions. This observation suggests that the promoting effect
96
of the metal oxides in the hierarchical catalysts is largely suppressed by SO2 at 450 °C, while the
active phase can be largely regenerated.
Figure 3.32
Methane Catalytic oxidation: Effect of 50 ppm SO2 dry aging for 30 min at
different temperatures on the catalytic activity of Pd@CeO2 (orange line),
Pd@ZrO2 (blue line), Pd@CZ (red line) and Pd/Si-Al2O3 (black line).
Conditions: 0.5% CH4; 2% O2; 50 ppm SO2 (if present) Ar balance,
GHSV=200000 mLg-1h-1.
To look for longer-term effects, the methane conversions under dry conditions at 500 °C
were monitored over each of the catalysts for 2 h during SO2 exposure (Figure 3.33). The
hierarchical catalysts deactivated sharply during the first hour of treatment, reaching a plateau at
around 60 % of the initial conversion. The Pd@ZrO2 and Pd@CZ catalysts showed similar
deactivation trends and were partially reactivated over time after reaching that plateau. The
Pd@CeO2 underwent a slightly slower deactivation but was not reactivated after reaching the
plateau.
97
Figure 3.33
Methane Catalytic oxidation: SO2 dry aging at 500 °C. Pd@CeO2 (orange line),
Pd@ZrO2 (blue line), Pd@CZ (red line) and Pd/Si-Al2O3 (black line) Conditions:
0.5% CH4; 2% O2; 50 ppm SO2, Ar balance, GHSV=200000 mLg-1h-1.
Aging at 500 °C using different SO2 concentrations resulted in different deactivation rates
but did not significantly change the plateau conversion (not shown). The residual conversion is
likely associated with the partially poisoned Pd phase, unpromoted by the metal oxide. Indeed, the
Pd/Si-Al2O3 sample was very stable under SO2 aging at 500 °C, indicating that the SO2 effect on Pd is
largely inhibited at this temperature. Moreover, MOx/Si-Al2O3 catalysts without Pd in their
formulation, prepared following a similar procedure as that used to synthesize the hierarchical
catalysts, did not exhibit any methane conversion at 500 °C and conversions on these materials did
not light-off until above 550 °C (Figure 3.34).
98
Figure 3.34
Methane oxidation light-off curves over CeO2/Si-Al2O3 (orange) and
Ce0.6Zr0.4O2/Si-Al2O3 (red curve). Conditions: 0.5% CH4, 2.0% O2, Ar balance,
GHSV=200000 mLg-1h-1, heating and cooling rates 10 °Cmin-1.
In parallel experiments, consecutive 50 ppm SO2-aging treatments for periods of 30, 60 or
120 min at 500 °C caused partial but irreversible deactivation of the Pd@CeO2 and Pd@CZ samples
(Figure 3.35). Conversely, the poisoning effect of SO2 on Pd@ZrO2 and Pd/Si-Al2O3 catalysts was
reversible at this temperature (Figure 3.35), even after 12 h aging (data not shown). Heating the
Pd@ZrO2 catalysts that had been aged at temperatures below 500 °C in dry conditions also resulted
in complete reactivation of the materials. Finally, it should be noted that the conversion trends
reported in Figure 3.32 and Figure 3.35 were affected by chromatographic effects due to the time
required for the entire catalyst bed to be exposed to SO2 at these low concentrations. Therefore,
the conversion at the beginning of reactivation can only be extrapolated.
99
Figure 3.35
Methane Catalytic oxidation: reactivation trends following consecutive SO2
dry aging treatments at 500 °C. Pd@CeO2 (orange line), Pd@ZrO2 (blue line),
Pd@CZ (red line) and Pd/Si-Al2O3 (black line). Conditions: 0.5% CH4; 2% O2; 50
ppm SO2 (if present) Ar balance, GHSV=200000 mLg-1h-1.
At 600 °C, under dry conditions, methane was completely converted to CO2 over all the
catalysts and there was no loss of conversion during SO2 exposure. To increase the sensitivity of the
experiment, the catalysts were diluted by addition of Al2O3 in a ratio of 1:3 and aged using the
same conditions as those employed for the pure catalysts (Figure 3.36). The diluted Pd@MOx
catalysts all achieved complete conversion at 600 °C prior to SO2 exposure, but the diluted Pd/Si-
Al2O3 catalyst showed only 80% conversion under these conditions. Exposure of the Pd@CeO2 and
Pd@CZ samples to SO2 at this temperature caused the conversion to decrease continuously with
time; however, for the Pd@ZrO2 and Pd/Si-Al2O3 samples, after displaying an initially sharp decline
in conversion, conversions partially recovered and reached a plateau value. This behavior was
reproducible and was never observed for CeO2-containing catalysts. A possible explanation for the
transient behavior on Pd@ZrO2 and Pd/Si-Al2O3 is that SO2 initially deactivates the PdO active
100
phase, competing with CH4 for oxidation. Later, SO3 and sulfate species formed by oxidation
migrate to the promoter/support. It has been previously reported that the activity of Pd/ZrO2
catalysts can increase after SO2 aging due to formation of a composite site between PdO and
sulfates at the PdO-support interface(151). It is noteworthy that there was also an increase in
conversion with SO2 exposure time at 500 °C for Pd@ZrO2 and Pd@CZ catalysts (Figure 3.33),
although the time scale for increasing conversion was much larger and the initial dip was not as
sharp.
Figure 3.36
Methane Catalytic Oxidation: SO2 dry aging at 600 °C and reactivation
trends. Pd@CeO2 (orange line), Pd@ZrO2 (blue line), Pd@CZ (red line) and
Pd/Si-Al2O3 (black line). Conditions: 0.5% CH4; 2% O2; 50 ppm SO2 (if present)
Ar balance, GHSV=200000 mLg-1h-1.
101
3.5.3 Model Catalysts
Photoelectron spectroscopy (PES) is a very useful technique to investigate the surface
chemistry of materials, in terms of chemical species accumulation/desorption, change in chemical
state of atoms, and surface/bulk distribution of species. However, insulating materials such as the
high surface area catalysts studied in this work are not suited for PES analysis, because of dramatic
charging effects that cause inhomogeneous shifts in the spectrum signals. For this reason we
developed model catalysts suitable for photoelectron spectroscopy analysis (i.e. having a
conductive support) designed to resemble the real catalysts in terms of surface composition and
resistance to high temperature oxidizing treatments (calcination/aging).
The choice of a proper support was a rather difficult task: aluminum and Al-alloys were
tested in the first place, but were discarded due to the low melting points and uncontrolled growth
of thick Al2O3 layer (leading to loss of surface conductivity). Stainless steel supports covered with
thin Al2O3 layers deposited by ALD were also tested, but the difference in thermal expansion
coefficients of steel (12-16x10-6 °C-1) and alumina (8.1x10-6 °C-1) resulted in cracking and leaching of
Al2O3 and even its incorporation in thermally grown Fe, Mn and Cr mixed oxides. Thermal oxidation
and expansion problems were finally bypassed by the choice of a conductive oxide, i.e. Indium-Tin
Oxide (ITO), supported on quartz for mechanical stability. ITO is stable in air over 900 °C without
experiencing loss of conductivity and has a thermal expansion coefficient very similar to that of
Al2O3 (8.1-8.6x10-6 °C-1).
On the basis of these preliminary observations, the supports for the model catalysts were
prepared by ALD of Al2O3 overlayers of various thicknesses (2, 5 and 10 nm) on an ITO/quartz
support (flat, low-surface-area material). The Al2O3 layer deposition was also carefully optimized,
because it caused a drop in surface conductivity as the thickness of the layer increased. 10 nm thick
layers were discarded because heavy charging was observed during preliminary XPS analysis. On
the other hand, ITO exposure after calcination had an opposite trend, increasing with decreasing
layer thickness. As revealed by Ion Scattering Spectroscopy (ISS), a 2 nm layer Al2O3 was not
enough to completely cover ITO, exposing it to the environment (Figure 3.37). 5 nm Al2O3 slides
were selected for further investigation because of minimized ITO exposure and negligible charging
during SRPES/XPS analysis.
102
Figure 3.37 ISS results for slides calcined at 850 °C for 5h: ITO/Quartz slides (black line), 2
nm ALD-Al2O3/ITO/quartz (red line), 5 nm ALD-Al2O3/ITO/quartz (green line).
The structure of the model catalysts used in this study is depicted graphically in Figure 3.38.
A 1-mm thick quartz slide was used to support a 500-nm thick conductive layer of ITO, over which a
5 nm Al2O3 layer was grown by ALD. To obtain a silanized, hydrophobic Si-Al2O3 comparable to that
of the HSA Si-Al2O3, the slides were allowed to react with TEOOS by soaking them in TEOOS
solutions, diluted with toluene, for 2 days. Finally, the slides were loaded with Pd@CeO2, Pd@ZrO2
or [email protected] (Pd@CZ) particles and calcined at 850 °C for 5 h. The catalysts are labeled
Pd@CeO2-m (m for model), Pd@ZrO2-m and Pd@CZ-m, respectively.
103
Figure 3.38
Graphical representation of Pd@MOx/Si-Al2O3/ITO/quartz model (A, not to
scale); AFM 3D image and mapping of ITO/quartz (B1), ALD-Al2O3/ITO/quartz
(B2), Pd@CZ-m calcined at 850 °C for 5 h (B3). EDS wide spectrum of calcined
Pd@CZ-m (C) and SEM image of Pd@CZ-m with 1 m view field (D1) and 10
m view field (D2).
AFM analysis of the model catalysts was performed at different stages in the preparation
procedure (Figure 3.38B). The bare ITO surface shows characteristic worm-like structures (Figure
3.38B1) which become slightly more defined in shape after Al2O3 deposition (Figure 3.38B2). The
final structure obtained after depositing Pd@MOx, followed by calcination at 850 °C, shows a similar
morphology, but with some aggregates on the surface (In Figure 3.38B3, see the white spots in the
2D image, blue in the 3D image). Similar to the results reported in a recent paper by Zhang et al.
(150), SEM/EDS analysis revealed aggregates composed of Pd with CeO2, ZrO2 or CZ on top of small
features that covered the Al2O3 layer (Figure 3.38D). A representative EDS spectrum of Pd@CZ-m is
reported in Figure 3.38C. The spectra obtained from different spots on the model catalysts revealed
104
a uniform composition with the expected Pd:(Ce+Zr) and Ce:Zr molar ratio (1:5.5 and 6:4
respectively). CeO2- and ZrO2-based catalysts also showed the expected Pd:MOx molar ratio. No
differences in surface morphology were observed for CeO2, ZrO2 or CZ-containing samples.
Since the SRPES signals for Pd 3d and Zr 3p overlap (3p1/2: 343eV, 3p3/2: 330 eV), only Pd 3d
spectra of Pd@CeO2-m samples are reported in Figure 3.39. Calcination temperature had a
significant effect on the Pd 3d region. Reference Pd@CeO2-m samples calcined at 500 °C showed
the Pd spin-orbit split doublet (Pd 3d5/2 and Pd 3d3/2) at 337.2 eV and 342.2 eV (Figure 3.39, A1),
which can be assigned to PdO, based on previous XPS and X-ray Absorption Near Edge Structure
(XANES) analysis of similar materials (152). A shift of the PdO signal to higher BE (almost 1 eV) with
respect to a bulk value is likely due to a size effect (129 135). Calcination of the Pd@CeO2-m
samples to 850 °C causes partial agglomeration of some of the nanostructures to form Pd and CeO2
particles in the range of 50 to 100 nm (Figure 3.39, B2) (150); and this agglomeration gives rise to an
additional contribution in the Pd 3d spectra, more typical for bulk PdO (366.5 eV) (Figure 3.39, B2).
Figure 3.39
Left: Pd 3d SRPES spectra at different excitation energies for Pd@CeO2-m
calcined at 500 °C (A1) and 850 °C (B1). 500 eV: red line, 640 eV: green line,
880 eV: blue line. Grey lines are guidelines to the eye. Right: representative
SEM images of Pd@CeO2-m samples calcined at 500 °C (A2) and 850 °C (B2);
bars: 200 nm.
105
Because the relative contribution of the larger agglomerates in photoemission is expected
to increase with excitation energy, we examined the surface:bulk signal contribution by tuning the
SRPES excitation energy. In agreement with this, the signal at higher BE, tentatively assigned to
PdO NPs, was relatively more intense at lower excitation energies, implying that this feature is
associated with a considerably smaller structures. An alternative explanation is that the two
contributions to the SRPES signal could be due to a Pd-Ce mixed oxide and a bulk PdO. For
example, the Pd 3d signal of Pd Ce mixed oxides has been reported to have BE of 338 eV BE (153).
Since the mixed oxides would likely segregate at higher temperatures, resulting in bulk PdO
formation, this could also explain the two features in the spectra (154). However, we regard this
explanation as less probable since the synthesis of the core-shell materials starts with pre-formed
Pd nanoparticles and complete disruption of the PdO and incorporation into CeO2 seems unlikely
to occur at 500 °C. Also, this latter explanation does not agree with the observation of a PdO (101)
reflection in the HSA catalysts calcined at 850 °C (Figure 3.26) or with previous EXAFS studies on
similar systems (89, 152). Due to the lower resolution, only one contribution averaging the signals
associated to PdO NPs and aggregates was observed in XPS (Figure 3.40).
In order to study the effect of SO2 aging at different temperatures, the model catalysts were
treated under similar conditions to those used with the HSA catalysts (except for higher GHSV) and
then transferred under inert atmosphere to the analysis chamber for analysis by XPS/SRPES. First,
the catalysts were aged for 2 h under dry methane-oxidation (0.5% CH4, 2.0% O2) conditions with
50 ppm SO2 at 350, 500 or 600 °C. Then, after XPS analysis, the aged catalysts were regenerated
under dry methane-oxidation conditions, without SO2, for 2 h at the same temperature at which
they were aged. After SO2 aging at 350 °C, the XPS spectrum showed a peak at 335.2 eV, which can
be assigned to metallic Pd (Figure 3.40) (71, 155, 156). Partial reduction of PdO could result from SO2
oxidation to SO3 by PdO lattice oxygen. The complete re-oxidation of the then formed metallic Pd
would not be possible under the conditions of this study at 350 °C, as previous XANES results on
HSA Pd@CeO2/Si-Al2O3 suggest, because of the low temperature and oxygen pressure involved
(89). A similar effect has been reported by Venezia et al. after aging Pd-based catalysts on TiO2 or
SiO2 at the same temperature overnight (71, 155) and by Liotta et al. for Pd/CeO2 catalysts operated
under lean conditions and 10 ppm of SO2 (156).
Along with the reduction of the Pd, there was a slight increase of the XPS signal at higher
BE following SO2 aging at 350 °C. The shift of the signal is less than that expected for Pd in the form
of PdSO4 and we suggest that this might be indicative of some interaction with sulfates groups, in
agreement with the literature (23, 84). After 2h under dry methane-oxidation conditions at 350 °C,
106
the XPS spectra did not change appreciably from those of the SO2 aged catalyst. An irreversible
poisoning of the PdO active phase would explain the similar behavior observed for HSA catalysts,
regardless of the presence and composition of the promoter (Figure 3.32). At higher temperature
(500 °C to 600 °C) SRPES spectra of the Pd 3d region did not change after SO2 aging or
regeneration, suggesting that the interaction of PdO with sulfates is weaker and reversible.
Accordingly, the Pd/Si-Al2O3 catalyst was very stable during dry SO2 aging at 500 °C (Figure 3.33)
and a sharp transient deactivation at 600 °C could only be observed on the diluted catalyst (Figure
3.36). These observations are in accordance with thermodynamic calculations on PdSO4 formation
from mixtures of O2 and SO3, predicting PdSO4 decomposition below 400 °C (157).
Figure 3.40 XPS spectra of Pd 3d region of Pd@CeO2-m, fresh (black line) and 350 °C SO2
aged (red line).
Independent of the SO2 aging temperature, the S 2p region of the SRPES/XPS spectra
showed only a signal corresponding to sulfate species (168-172 eV BE). The use of SRPES was
crucial in order to observe an intense and resolved S 2p signal, thanks to the much higher
photoionization cross-section (0.83 Mbarn at 425 eV; 0.022 Mbarn at 1486.6 eV) (158) and shorter
information length, which favors signal from surface species. Indeed, after 500 °C aging, sulfates
were only observed in SRPES spectra and not in XPS spectra (Figure 3.41A,C), which also showed an
overlapping Zr 3d signal due to Al K (1476.8 eV) and K (1474.8 eV) lines. After 2 h of
regeneration at 500 °C, sulfates were partially removed from Pd@ZrO2-m, while the sulfates
remained on the Pd@CeO2-m and Pd@CZ-m samples, a finding consistent with the regeneration
trends observed with the HSA catalysts (Figure 3.34).
107
During regeneration of the catalyst, only a small fraction of sulfur species desorbed,
suggesting that a rearrangement of sulfates on the surface could lead to reactivation and that TPD
studies may give misleading results for evaluation of SO2 poisoning resistance. At 600 °C, more
sulfates are formed (even if a decrease in conversion was observed only in diluted samples) and
these could be observed both in SRPES and XPS spectra (Figure 3.41B,C). Based on the cross section
corrected photoemission signal of S and Ce (and/or Zr) in XPS, we estimate a MOx:S molar ratio of
5:1 for CeO2 and CZ and 7:1 for ZrO2. These ratios are consistent with about 1 monolayer (ML) of
sulfate being formed on the surface of the oxide nanoparticles, considering that the XPS
information depth is around 2 nm, which corresponds to about 6 ML of CeO2 in the close-packed
(111) direction. Regeneration under dry conditions at 600 °C resulted in desorption of sulfates from
the surface/subsurface of all the samples (Figure 3.41B), consistent with similar trends observed in
the recovery of conversion (Figure 3.36).
Figure 3.41
S 2p region of SRPES spectra of SO2 aged and regenerated samples at (A)
500 °C and (B) 600 °C (orange: Pd@CeO2-m; red: Pd@CZ-m; blue: Pd@ZrO2-m;
thick line: aged; thin line: regenerated); (C) S 2p region of XPS spectra of SO2
aged samples at 500 °C (red line) and 600 °C (green line) (top: Pd@CeO2-m;
middle: Pd@CZ-m; bottom: Pd@ZrO2-m).
Previous FTIR (145), TPD (142) and XPS (159) studies reported sulfate desorption from
Ce1Zr1-xO2 in the range of 600 °C to 700 °C. The desorption of sulfate species from pure ZrO2-based
materials is much less investigated, with TPD results showing higher desorption temperatures than
for CeO2 sulfates (800 °C) (142). We suggest that the surface sensitivity of SRPES/XPS analysis
108
allowed us to observe a partial desorption of sulfates from ZrO2, occurring at lower temperature,
which could not be detected by TPD. The difference of observed sulfate stability could also be due
to the atmosphere in which desorption was performed, which can affect decomposition of S-
species (138). Indeed, Colussi et al. observed different TPD profiles on SO2-aged Pd/CeO2/Al2O3
samples when the desorption was carried out in reaction conditions or inert atmosphere (138),
suggesting a role of in-situ produced water in the SO2 desorption mechanism.
Interestingly, the BE for the S 2p features in Figure 3.41 was different for the Pd@CeO2-m
and Pd@ZrO2-m samples, implying that sulfates bonded to Zr or Ce cations are distinguishable.
Since the spectra of Pd@CZ-m are very similar to that of Pd@CeO2 both in shape and
formation/desorption trends, it appears that primarily Ce sulfates are formed on Pd@CZ-m samples
and that these are similar to sulfates formed on CeO2. In agreement with this, the Ce 3d and Zr 3d
XPS spectral regions show modifications after SO2 aging that depend on the MOx composition
(Figure 3.42).
Figure 3.42
XPS spectra of pristine samples (black line) and SO2 aged samples at 500 °C
(red line) and 600 °C (green line). (A) Ce 3d region of Pd@CeO2-m; (B) Zr 3d
region of Pd@ZrO2-m; (C1) Ce 3d region and (C2) Zr 3d region of Pd@CZ-m.
109
In particular, Ce 3d spectra of Pd@CeO2-m (Figure 3.42A) and Pd@CZ-m (Figure 3.42C1)
aged at 600 °C are similar and exhibit more intense Ce(III) contributions than what is observed on
the fresh catalyst, probably due to reduction of CeO2 by SO2 with consequent formation of cerium
sulfate. By contrast, Zr 3d spectra for Pd@ZrO2-m (Figure 3.42B) are different from spectra for
Pd@CZ-m (Figure 3.42C2), revealing sulfate formation on Pd@ZrO2-m after aging at 500 °C and 600
°C, while the Pd@CZ-m spectra do not change. This is consistent with the formation of sulfates only
on Ce cations in CZ materials, which leaves the chemical environment of Zr cations unaffected.
Figure 3.43 shows the O 1s spectra of the fresh samples, exhibiting peaks typical of bulk
oxygen (529 eV for CeO2 and 531 eV for ZrO2) and hydroxyl species (532 eV) (127). SO2 aging leads
to an increase in the intensity of the higher BE contribution (532 eV) associated with OH groups
(Figure 3.43). These results, taken together with the trends observed in S 2p spectra, suggest that
the variation of O 1s signals is mostly due to oxygen from sulfate species.
Figure 3.43
O 1s region of SRPES spectra of SO2 aged and regenerated samples at (A)
500 °C and (B) 600 °C (orange: Pd@CeO2-m; red: Pd@CZ-m; blue: Pd@ZrO2-m;
thick line: aged; thin line: regenerated); (C) O 1s region of XPS spectra of SO2
aged samples at 500 °C (red line) and 600 °C (green line) (top: Pd@CeO2-m;
middle: Pd@CZ-m; bottom: Pd@ZrO2-m). SRPES and XPS spectra of fresh
samples are reported for reference (black lines).
110
3.5.4 Discussion
First, the observations here demonstrate that the previously reported synthesis of Pd@CeO2
and Pd@ZrO2 particles by self-assembly can be extended to form materials with mixed-oxide shells,
Pd@CexZr1-xO2. The pre-formed core-shell particles can be deposited on both high-surface-area
(HSA), functionalized alumina or on model Si-Al2O3 surfaces to achieve a similar, controlled
morphology. In all of the materials, the oxide shell was found to promote catalytic activity for
methane oxidation. The series of hierarchically structured catalysts supported on Si-Al2O3 also had
comparable surface areas and active-phase accessibilities. These properties make them good
candidates to study the effect of the oxide promoter composition on SO2 poisoning.
The methods used in preparing the model catalysts in this study are also of general interest
for investigating the effects of temperature and operating conditions. The approach is based on
the ability of ALD to form uniform, thin films having precise thicknesses and a wide range of
possible final compositions. Indeed, model supports could be prepared from almost any metal
oxide by varying the ALD precursors(160). This allows support composition to be varied
systematically in order to study the effect of metal-support interactions on the catalyst
performance, morphology (by SEM) and chemical state (by XPS/XANES). Also, model catalysts
prepared in this way can be tested and analyzed repeatedly, making it possible to study the effect
of aging and regeneration treatments together with properties for adsorption, desorption or
redistribution of surface species. The lower surface area of model catalysts, that affects the range of
achievable GHSV, is the only drawback of this powerful method of investigation.
SO2 poisoning of Pd-based catalysts in lean catalytic oxidation of methane proceeds via
two distinct mechanisms, depending on the aging temperature. At lower temperature (<450 °C),
SO2 irreversibly poisons PdO, causing a partial reduction to Pd and the formation of sulfates in close
proximity to the active phase. The temperature is not high enough to allow the spillover of sulfate
species from the active phase to the support/promoter and to promote a complete re-oxidation of
the poisoned catalyst (89, 161). At higher temperatures, PdO is more resistant to poisoning, thanks
to the spillover of sulfates from the active phase and/or the direct reaction of SO2 with the
promoter. However, since the promoting effect of the metal oxide is turned off by sulfate
poisoning, the catalysts deactivate until a plateau is reached, corresponding to the active phase
residual activity.
The effect of the oxide promoter composition on SO2 poisoning is also temperature
dependent. At 500 °C, Pd@ZrO2 catalysts showed the best sulfur resistance among the samples,
being only reversibly deactivated thanks to partial sulfate desorption. The performance of Pd@CZ
catalysts is intermediate between that of the two pure oxides, suggesting that no cooperative
111
effect of Ce and Zr takes place. On the other hand, at 600 °C, Pd@CeO2 and Pd@CZ can stabilize the
catalyst against transient deactivation, acting as a sink for SO2, while Pd@ZrO2 deactivates in a
similar way as Pd/Si-Al2O3. However, the poisoning effect of SO2 is less relevant than at 500 °C, since
the deactivation is observed only for very high GHSV. These results make Pd@ZrO2 catalysts a more
suitable candidate for real application, taking into account the better stability in the presence of
H2O with respect to Pd@CeO2 catalysts.
The chemical-state and surface sensitivity of photoelectron spectroscopy also allowed for
the observation of preferential sulfation of Ce over Zr in CZ mixed oxides. The chemical
environment of Zr cations only changed after sulfation of ZrO2, while in the case of CZ no evident
differences were observed. These results strongly suggest that sulfate species are associated with
individual metal cations, producing a first evidence of what was tentatively proposed by Luo et al.
based on TPD and pulse-reactor experiments (142).
3.5.5 Conclusions
The self-assembly methodology described previously (162) was modified in order to
synthesize nanostructured Pd@CexZr1-xO2 (Pd@MOx) units in the whole compositional range
(0<x<1). The synthesis of dispersed Pd@MOx allowed the preparation of a series of high-surface-
area Si-Al2O3 supported catalysts and model catalysts having similar nanostructure and surface
chemistry. Comparison of results on the two types of catalysts allowed the SO2 poisoning of
methane oxidation on Pd-based catalysts to be systematically studied to elucidate the role of the
MOx promoter and the aging conditions. At lower temperatures (<450 °C), the PdO active phase is
irreversibly poisoned by SO2 due to interaction with sulfates which are not able to spillover to the
support/promoter. At higher temperatures (>500 °C), poisoning is slowed by formation of sulfate
species on the oxide promoter. Due to partial decomposition of sulfates at 500 °C, Pd@ZrO2-based
catalysts showed the best sulfur-poisoning resistance, attaining complete regeneration even after
prolonged aging, and thus they are the best candidates for real application. [email protected]
catalysts showed intermediate sulfur tolerance compared to Pd@CeO2 and Pd@ZrO2, in agreement
with previously reported results (142). The high chemical sensitivity of PES techniques provided
direct evidence for previously suggested formation of sulfate species on individual metal cations in
CexZr1-xO2 mixed oxides (142). Finally, the model-catalyst approaches developed here should allow
the study of metal-support interactions in other catalytically relevant systems by simply varying the
ALD-deposited thin film composition.
112
4. Biomass to Biofuels:
Hydrodeoxygenation Reaction
4.1 Introduction
Biomass is organic matter derived from living organisms, and it has been harnessed as an
energy source since the discovery of fire. Nowadays, it can either be used directly via combustion
to produce heat, or indirectly via thermal, chemical, or biochemical conversion to various biofuels
and chemicals. Depending on the origin and production technology of biofuels, they are generally
classified as first, second or third generation biofuels (27). First-generation biofuels are produced
directly from food crops by fermentation to bioethanol or by abstraction of oils for use in biodiesel.
However, such feedstock provide only limited biofuel yields and have a negative impact on food
availability and security. To overcome the limitations of first-generation biofuels, second-
generation biofuels have been developed. These are produced from feedstock of lignocellulosic,
non-food materials that include straw, forest residues and purposely grown energy crops on
marginal lands. The term third generation biofuels has only recently been introduced and it refers
to biofuels based on algal biomass production. Algae are capable of much higher yields than other
feedstock and can be genetically manipulated to produce a particular feedstock. Most notably, it
has been suggested that algae might be tied directly to carbon emitting sources (power plants,
industry, etc.) where they could convert emissions into usable fuel, thus dramatically improving the
plants sustainability.
A wide range of chemical processes may be used to convert lignocellulosic biomass into
more conveniently used, transported or stored biofuels (Figure 4.1). Gasification, pyrolysis,
hydrolysis and biochemical routes are the main strategies to transform lignocellulosic biomass in
intermediate products that can be further processed to produce valuable chemicals or fuels.
Gasification produces syngas (CO + H2), containing traces of methane, also called a producer gas,
suitable for internal combustion engines or further processing to produce methanol, alkanes (via
Fischer-Tropsch synthesis) or hydrogen (163). Pyrolysis and liquefaction of lignocellulosic biomass
yields bio-oils that can be upgraded to yield liquid fuels. Conversion of biomass to biofuels can also
be achieved via selective conversion of individual components of biomass. For example, cellulose
and hemicellulose can be converted over solid acid catalysts via hydrolysis to intermediate
platform chemicals such as sorbitol (164), glucose, xylose, levulinic acid (165, 166), 5-
113
hydroxymethylfurfural (HMF) (167), and furfural (168), to name a few. Among the many upgrading
processes of lignocellulosic feedstock, hydrodeoxygenation (HDO) is a pivotal reaction to produce
liquid fuels (163, 169).
Figure 4.1 Main strategies for fuel production from lignocellulosic biomass.
HDO is a hydrogenolysis process for removing oxygen from oxygen-containing
compounds. It is a very attractive process due to high carbon efficiency (100% theoretical) and
technology compatibility with existing petroleum hydrotreating technology. HDO requires
relatively high H2 pressure (7-40 bar), but in principle H2 can be sustainably produced from biomass
by sugar aqueous reforming, alcohol steam reforming, syngas steam reforming or using solar or
wind power. In general, biomass-derived molecules contain many different functional groups,
which are hydrogenated at different temperatures and may or may not be selectively activated by
a catalyst (29). Moreover, the composition and oxygen content of bio-oils and hydrolyzed cellulosic
feedstock varies dramatically depending on the source, and this results in different HDO reactivity.
Therefore, the main strategy to study the performance of a certain catalyst in HDO reaction is to
study model compounds activation. By understanding the chemistry and the reaction mechanisms
of a certain model compound, one can study the selectivity to certain products and get insights for
upgrading real feedstock to desired fuels and chemicals. In the present work, we study HDO of two
very important model molecules: furfural and HMF.
114
Furfural is formed during pyrolysis of biomass (170, 171) and can be obtained in high yields
from dehydration of C-5 sugars, such as xylose and arabinose (172, 173). By selective HDO, furfural
produces 2-methylfuran (MF) (174), which can be used as gasoline additive due to its high octane
rating and low solubility in water. HMF is another key platform chemical in biomass conversion,
readily obtained by the acid-catalyzed dehydration of C-6 sugars (174 178). The selective HDO
reaction of HMF leads to 2,5-dimethylfuran (DMF), which has a high energy density and an octane
rating of 119 and can be used as fuel additive (179). Moreover, both MF and DMF can be reacted
with ethylene via Diels-Alder reaction to yield toluene (180, 181) and p-xylene, respectively (181).
Despite the recent advances in catalytic HDO reaction of HMF and furfural, some critical
issues remain to be addressed, among which the understanding of the factors influencing high
selectivity to MF and DMF and catalytic stability. In the present chapter, we aim to discuss the
mechanism of HDO of furfural and HMF in pressurized flow reactors and to rationalize the influence
of H2 pressure and bimetallic catalysts nanostructure on the catalytic HDO reaction.
115
4.2 H2 Pressure Dependence of HDO Selectivity for Furfural over Pt/C Catalysts
4.2.1 Introduction
Furfural has a relatively high vapor pressure at low temperature, so that HDO reaction of
furfural can be performed in simple, vapor-phase reactors, and much of the published work in this
area has been performed in this way (38, 182, 183). Decarbonylation to form furan is the major side
reaction for essentially all metal catalysts and significant effort has gone into finding materials that
show higher selectivity to MF, rather than producing furan (38, 182 184). Interestingly,
decarbonylation is not reported to be a major side reaction in HDO studies of 5-
hydroxymethylfurfural (HMF) (41, 185), which has a very similar structure to furfural. A major
difference between HMF and furfural is that HMF has a much lower vapor pressure, so that reaction
studies of HMF are commonly performed at high H2 pressures with most of the HMF in the liquid
phase, usually in the presence of a solvent (186 188). Recent theoretical studies have indicated that
the furfural adsorption energy and conformation on metal catalysts can be profoundly changed by
the hydrogen surface coverage, which in turn can affect the relative rates of hydrogenation and
decarbonylation (189). Therefore, the fact that decarbonylation is observed in HDO of furfural, but
not in HDO of HMF, may be due to differences in the reaction conditions.
In the present section, we examine the effect of reaction conditions on HDO of furfural over
a Pt/C catalyst. The high-pressure, three-phase HDO reaction of furfural was performed in a tubular
flow reactor described in detail in Section 2.10. Low-pressure reactions were carried out in the
same temperature range but at one atmosphere total pressure, with variable H2 partial pressures.
At high pressures, HDO of furfural was found to be very similar to HDO of HMF: decarbonylation
was negligible and the reaction was sequential, with all of the reactants proceeding through MF,
which in turn reacted to over-hydrogenated products. At lower H2 pressures, decarbonylation
occurred in parallel with HDO, and was monotonically suppressed with increasing H2 pressures.
4.2.2 Catalyst Synthesis and Characterization
A 10 wt % Pt/C catalyst was prepared by impregnation of carbon black (Vulcan XC-72R)
with a water/ethanol (5:1) solution of tetraammineplatinum(II) nitrate (Pt(NH3)4(NO3)2, 99.99%, Alfa
Aesar). The dried powders were heated in He flow at 500 °C for 6 h using a heating ramp of 3 °C
min-1 and then used directly in the reactor without further pretreatment. After reduction in H2 at
400 °C, the Pt dispersion was determined to be 12% by CO chemisorption and the Pt particle sizes
observed by TEM varied between 4 and 10 nm (Figure 4.2).
116
Figure 4.2 Representative TEM image of 10 wt% Pt/C
4.2.3 High Pressure HDO of Furfural
For work at high pressures, the reactor was part of the continuous flow system described in
Section 2.10. The liquid feed (1-wt% furfural in 1-propanol) flow rates were varied from 0.02 to 0.2
ml min-1, while the H2 flow rates were 2 to 20 mL min-1 (STP). The reactor was held at 180 °C and 33
bar total pressure in all measurements, varying only the flow rates of reactants while maintaining a
constant ratio of gas and liquid flow rates. The environment in this system is similar to that of a
trickle-bed reactor and mass transfer processes, including diffusion of furfural, H2, and products
within the liquid phase and within the solid catalyst, are complex. Quantitative analysis of the
reaction rates for such system is difficult, since the rates are dependent on catalyst wetting and
mass transfer. The carbon balance (based on furfural) was always better than 90% in both high-
and low-pressure measurements.
Figure 4.3 shows the furfural conversion and the product yields as a function of the reactor
space time, W/F, defined here as the weight of the catalyst divided by the volumetric flow rate of
the liquid. A guide to the products that are formed is shown in Scheme 4.1.
117
Figure 4.3
Conversion and product distribution for the liquid-phase HDO reaction of
furfural over a 10 wt% Pt/C catalyst as a function of reactor space time. (A)
the overall product distribution; (B) over-hydrogenated products in detail
(product group D). Reaction conditions: 33 bar and 180 °C. () furfural
conversion, () product group B, ()MF, () product group D, () MTHF,
() 2-pentanol, () pentanedione, ()2-propoxypentane.
Scheme 4.1 Reaction network for liquid-phase hydrodeoxygenation of furfural in alcohol
solvent.
118
The results in Figure 4.3A correspond very closely to what was reported earlier for HDO of
HMF for the same catalyst and reaction conditions (185). The furfural conversion increased with
residence time, reaching 100% conversion for the highest W/F values. At the shortest residence
times, the main products were 2-methylfuran (MF) and several partially hydrogenated products
designated as B (furfuryl alcohol ─ FA, furfuryldipropyl acetal ─ FAct, and furfurylpropyl ether ─
FEther). At such short residence time, there was no evidence for over-hydrogenated products,
listed as D in Scheme 4.1. With increasing residence time, the yield of B products decreased steadily
while the yields of MF and D products increased (Figure 4.3B). With still longer residence times, the
MF yield reached a maximum and began to decrease, while the D products continued to increase.
Several important observation can be made about the results presented in Figure 4.3. First,
there was no evidence for decarbonylation products. No furan was formed and there were no C-4
products in the final D products. Second, as previously reported for HDO of HMF, the evolution of
products indicates that HDO of furfural at high pressure is a series reaction, as shown in Scheme
4.1. This is demonstrated by the following facts: 1) The partially hydrogenated, B products form first
and decrease continually with residence time. 2) The MF yield goes through a maximum, indicating
it is formed from B, then declines. 3) The over-hydrogenated D products are absent at short
residence times and increase continually.
As shown in Figure 4.3B, the major over-hydrogenated product is 2-methyltetrahydrofuran
(MTHF). The fact that MTHF is formed only after MF has formed implies that ring hydrogenation can
only occur after reduction of the carbonyl group of the furfural. Other products that formed in
parallel with MTHF were 2-pentanol, pentanedione, and 2-propoxypentane (a reductive-
etherification product formed by reaction of 1-propanol with 2-pentanone).
As a further test of the sequential reaction network proposed in Scheme 4.1, experiments
were carried out in which the intermediate reactants were fed to the reactor. Figure 4.4 shows the
MF conversion and final product distribution as a function of space time, at 180 °C and 33 bar,
when 0.8 wt% MF in 1-propanol was fed to the reactor. The molar concentration of MF in these
experiments was the same as that of furfural in Figure 4.3. The data in Figure 4.4 support the
proposed sequential reaction scheme. The conversion of MF increases with residence time and the
final product distribution is similar to that shown in Figure 4.3B. The slightly lower yields of
pentanedione, along with its etherification products, may be due to the fact that water is formed in
the reaction of furfural to MF and will therefore be present in lower amounts when MF is fed
directly to the reactor.
119
Figure 4.4
Conversion and product distributions for the reaction of MF in 1-propanol
over a 10 wt% Pt/C catalyst as a function of reactor space time. Reaction
condition: 33 bar and 180 °C. () MF conversion, () MTHF, () 2-pentanol,
() pentanedione, ()2-propoxypentane.
The reactivity of FA, MTHF, tetrahydrofurfuryl alcohol (THFA) and 2-pentanone were also
examined in less detail, with results shown in Table 4.1. For these experiments, the reactants were
diluted in 1-propanol to the same molar concentrations as furfural and their reactivity studied at
180 °C and 33 bar at a single reaction space time of 1 g min mL-1. In agreement with the data in
Figure 4.3 and Scheme 4.1, the conversion of FA under these conditions was 96% and the products
were similar to those observed from furfural, with the exception that FEther and FAct were not
observed. MTHF and THFA were completely unreactive under these conditions.
The low reactivity of MTHF indicates that ring-opening is difficult after the ring has been
saturated, in agreement with a previous report by Bradley et al. (190), who argued that the strong
adsorption of the aromatic ring on Group VIII metals can weaken the C-O bond in furan. The fact
that the hydroxyl group on THFA is also not reactive, whereas the hydroxyl group on FA is very
reactive, implies that the aromatic ring strongly affects the reactivity of that group. Finally, 2-
pentanone was completely converted at the investigated space time, with a 90% yield to the
reductive-etherification product, 2-propoxypentane. Only a small amount of 2-pentanol was
formed under these conditions. This result is consistent with the observations from the reactions of
furfural, explaining the presence of 2-propoxypentane as one of the major by-products.
120
Table 4.1 Product distributions of reaction from varied reactant over 10 wt% Pt/C, at
180 °C, 33 bar and space time of 1 g min mL-1.
4.2.4 Low Pressure HDO of Furfural
For the low-pressure measurements, furfural was introduced to the reactor by flowing a
carrier gas through a heated bubbler. In most cases, the partial pressure of furfural was maintained
at 0.02 bar. The total gas flow rate to the bubbler was fixed at 60 mL min-1, while the H2 partial
pressure was adjusted by mixing He with the H2. A bubble meter at the reactor exit was used to
continuously check the gas flow rates. The effluent from the reactor was diluted with 120 mL min-1
He to decrease the final concentration of furfural and avoid condensation problems. The product
analysis was carried out using a gas-tight syringe to inject the effluent from the reactor into the GC-
MS. Notably, in the high-pressure experiments, the conversion and selectivity were measured as a
function of the total flow rates in order to map out the product distribution as a function of
residence time. Since the reaction pathway was qualitatively different at the lower H2 pressures,
most of the low-pressure measurements were performed under differential conditions, with
furfural conversions less than 15%. Although the conversions increased from approximately 5 to
15% with increasing H2 partial pressure for these fixed flow rates, the selectivity did not change
significantly when the flow rates were adjusted to vary the conversions between 10 and 40%.
The selectivity for furfural HDO at low H2 pressures are shown as a function of H2 partial
pressure in Figure 4.5. Since furan, FA, and MF accounted for more than 95% of the products under
these conditions, only the selectivity to these products is shown, which is qualitatively different
from what observed at higher pressures. Consistent with observations from a large number of
literature studies on various metal catalysts, decarbonylation of furfural to furan was a major
reaction pathway in all cases in these low-pressure experiments (38, 182 184). At lower H2
pressures, the decarbonylation product, furan, was clearly the dominant product, with a selectivity
to furan of more than 70% at 0.1 bar H2. With increasing H2 pressure, the selectivity to furan
decreased and the selectivity to FA increased. At 1 bar H2, FA was the largest product.
121
Figure 4.5
Products selectivity for the vapor-phase HDO reaction of furfural over a 10
wt% Pt/C catalyst as a function hydrogen pressure by varying H2: He ratio.
Reaction conditions: total flow rate 60 mL min-1, furfural partial pressure 0.02
bar, total pressure 1 bar and 180 °C. () furan, () FA, () MF.
To rule out furan production from FA, experiments were conducted in which FA was fed to
the reactor. For these conditions, no furan was produced from FA and the major product was MF
(>60% selectivity). Additional products formed from FA were MTHF, THFA, and 2-pentanone. Since
all of these are 5-carbon molecules, there is no additional decarbonylation once FA is produced.
Therefore, the relative importance of decarbonylation and HDO in the reaction of furfural can be
determined from the furan selectivity in Figure 4.5.
The results in Figure 4.5 demonstrate the importance of H2 pressure in determining
decarbonylation selectivity over Pt/C catalysts. For the 33 bar experiment, we estimate that the H2
partial pressure was approximately 22 bar and the furfural pressure 0.14 bar, with the balance
being 1-propanol. Although the presence of 1-propanol may also affect the selectivity, an
extrapolation of the data in Figure 4.5 to higher pressures suggests that the higher H2 pressure is
the most likely reason for the absence of decarbonylation at 33 bar.
122
4.2.5 Discussion
The present results demonstrate the importance of H2 pressure in the reaction network for
furfural. For a typical Pt/C catalyst, decarbonylation to furan dominates at the lower H2 pressures
that are often used in laboratory testing of furfural HDO; however, a sequential reaction network, in
which all the furfural proceeds through MF, dominates at higher pressures. This may help unify the
literature for HDO of furfural and 2-hydroxymethyl furfural (HMF). Decarbonylation is not a major
pathway in most studies with HMF. Because of the low vapor pressure of HMF, catalyst testing for
HDO of HMF is usually performed in an autoclave at high pressures, where decarbonylation would
also not be important for the reaction of furfural.
The fact that decarbonylation is unimportant at higher H2 pressures agrees with previous
papers where HDO of furfural was studied in an autoclave at higher pressures (191, 192). What the
present work demonstrates is that the absence of decarbonylation in previous high-pressure
studies is probably not due to the catalysts that were studied but rather the conditions used to
study the reaction. In other words, the selectivity of a catalyst can be changed by simply varying
the H2 pressure.
The reason for the strong H2 pressure dependence on the reaction products is likely due to
changes in the hydrogen surface coverage. DFT calculations of furfural adsorption and reaction on
Pd(111) (189) have shown that the presence of co-adsorbed hydrogen changes the conformation
of adsorbed furfural. Furfural prefers to lie flat on Pd surface in the absence of adsorbed hydrogen
but the ring becomes perpendicular to the surface when hydrogen is present. Moreover, kinetic
modeling indicates that the selectivity of the hydrogenation product (FA) increases relative to the
decarbonylation (furan) with increasing hydrogen coverage. These calculations are in very good
agreement with the trends observed in the present low-pressure HDO study.
At high H2 pressures, the product distributions for the reactions of furfural and HMF are
very similar. In both cases, the reactions are sequential, proceeding through either MF or DMF. For
HDO of HMF, this was shown to be true for carbon-supported Pt, Pd, Ir, Co, Ni, and Ru (185). The
final products formed from the reaction of MF and DMF are also similar. In earlier work with HMF, it
was shown that Pt gives a mixture of products consisting of primarily open-ring ketones and
alcohols, with some of the ring-saturated dimethyltetrahydrofuran (185). With furfural, the ring-
saturated product, MTHF, is slightly favored but the product slate is similar. In both cases, the
production of open-ring products occurs only after reduction of the carbonyl group to a methyl
group.
123
The similarities in the reaction networks for furfural and HMF imply that the strategies for
developing improved catalysts should also be similar. This leads to the important question of
whether catalysts developed to minimize decarbonylation in the reaction of furfural will also be the
most selective for producing DMF from HMF. Based on earlier theoretical studies, it seems likely
that there will indeed be a strong relationship between minimizing decarbonylation at low
pressures and selectivity towards DMF at high pressures but a careful examination of the
relationship is needed. In both cases, bimetallic catalysts have been shown to exhibit improved
selectivity than their single-component analogues (38, 188, 193 195).
Most high-pressure HDO studies employ a solvent, which can affect the product
distribution. In addition to the fact that the solvent may also be a reactant (e.g. 2-propoxypentane
is a product of a reaction between 1-propanol and 2-pentanone in the present study), adsorbed
solvent molecules can also change the conformation of the adsorbed reactants in a manner similar
to that with adsorbed hydrogen. Because the chemistry in earlier studies with HMF performed
under the same conditions appeared to be similar for different solvents (41), we do not believe the
presence of the solvent can explain the differences observed in the high-pressure and low-pressure
results of the present study.
4.2.6 Conclusions
H2 pressure plays an important role in determining the product distribution in HDO of
furfural. Decarbonylation is a major reaction pathway only at lower H2 pressures. At higher H2
pressures, HDO of furfural follows a sequential reaction network, with MF formed as an
intermediate. Great progress is being made in our understanding of HDO catalysis for furfural and
HMF. The development of unifying concepts for different reactants and reaction conditions further
helps our understanding of these reactions.
124
4.3 Mechanisms for High Selectivity in HDO of HMF over Pt-Co Nanocrystals
4.3.1 Introduction
The reaction of HMF to DMF has been extensively studied over various metal and metal-
alloy catalysts (196 198), but selectivity to DMF over ring-opened (e.g. 2-hexanone, 2,5-
hexanedione) and ring-hydrogenated (e.g. 2,5-dimethyltetrahydrofuran ─ DMF) products is often
poor. Understanding what is required for a good catalyst is made more difficult by the different
yields that have been reported by different groups for materials of similar composition (186, 187).
Some of these variations appear to be due the sensitivity of results to the type of reactor that is
used in the rate measurements (41). Even so, bimetallic catalysts appear to be more selective than
their pure-metal analogs (38, 194, 195). For example, Schüth and co-workers (188) reported DMF
yields as high as 98% on Pt-Co nanoparticles which were encapsulated in hollow carbon spheres.
As already mentioned in the previous section, the reaction of HMF to DMF is sequential on
Pt and many other metals(185), with HMF reacting selectively to DMF, but DMF then going on to
form secondary products over the same catalysts. Based on these considerations, the yield of DMF
is determined by the relative rates of formation and consumption of DMF, and the most selective
catalysts should show low activity for reaction of DMF. An implication of this picture is that
selective catalysts cannot consist of a mixture of selective and nonselective components, since
even a relatively small fraction of nonselective material could convert DMF to other products. For
the case in which two metals are nonselective (e.g. Pt and Co)(185), but their alloy is selective, this
means that the composition of the metal particles must be uniform.
The synthesis of highly uniform Pt and Pt-Co nanocrystals (NCs) have been demonstrated
via solvothermal methods (39, 199). Solvothermal synthesis refers to a method which involves the
use of a solvent under moderate to high pressure (typically between 1 bar and 10,000 bar) and
temperature (typically between 100 °C and 1000 °C). The process can be used to prepare many
geometries including thin films, bulk powders, single crystals, and NCs. If properly optimized,
solvothermal synthesis can yield NCs having controlled and uniform size, morphology, and
composition, parameters that are strongly correlated to the NCs catalytic properties (39, 200).
Therefore, this synthetic approach allows to prepare and test catalysts whose properties are well
controlled, allowing us to understand the effect of various parameters on catalytic activities.
In this section we report the HDO of HMF into DMF catalyzed by Pt, Pt3Co, and Pt3Co2 NCs-
based catalysts using a continuous flow reactor. DMF yields as high as 98% were achieved with
Pt3Co2 alloys due to very low reactivity of DMF towards over-hydrogenated products. Furthermore,
the bimetallic catalysts synthesized using the solvothermal method are stable and superior to alloy
125
catalysts prepared by traditional wet impregnation. Characterization of the Pt-Co NCs revealed that
they have a special structure consisting of a monolayer of surface oxide on a metallic core.
Calculations using Density functional theory (DFT) rationalize the stability of this structure and
indicate that the oxide prevents side reactions while providing catalytic sites for effective
conversion of HMF to DMF. Computations reveal a radical-mediated reaction mechanism, which
was found to be critical for selective HDO on oxides. Structural characterization and DFT
calculations confirm that controlling the bimetallic composition is essential for preparing a good
catalyst.
4.3.2 Synthesis and Characterization of Pt- and Pt-Co-based Catalysts
Nearly monodisperse Pt, Pt3Co, and Pt3Co2 NCs were synthesized by using or modifying
reported methods (39, 199). The experimental setup used for the synthesis of NCs in this work is
schematically depicted in Figure 4.6. A Schlenk line connected to a vacuum pump was used to
operate under N2 inert atmosphere (1 bar, flow).
Figure 4.6 Schematic representation of the solvothermal synthesis setup used in this
work.
To prepare Pt NCs, 314 mg of platinum (II) acetylacetonate (Pt(acac)2, Acros, 98%) was
dissolved in 40 mL trioctylamine (TOA, Sigma-Aldrich, 97 %), 10.9 mL oleylamine (OAm,
SigmaAldrich, 70 %), 2.6 mL oleic acid (OAc, Sigma-Aldrich, 90 %), and 0.9 mL trioctylphosphine
(TOP, Acros Organics, 97 %). The reaction mixture was kept under vacuum at 80 °C for 30 minutes
and then heated up to 300 °C at a rate of 10 °C min-1. After 30 minutes, the reaction mixture was
cooled down to room temperature. Then 40 mL toluene were added. The mixture was divided into
126
6 centrifuge tubes (50 mL) and 30 mL isopropanol and 5 mL ethanol were added. After 2 minutes of
centrifugation at 6000 rpm, the supernatant was removed and the precipitate was re-dispersed in
hexane or toluene. After washing the excess amount of OAc, OAm, and TOA by isopropanol, the
NCs were dispersed in toluene.
To synthesize Pt3Co NCs, a method reported by Shevchenko et al. was scaled up(199). 264
mg of Pt(acac)2 was dissolved in 32 g hexadecylamine (Acros Organics, 90 %) and 16 mL diphenyl
ether (Sigma-Aldrich, 99 %) in the presence of 672 mg 1- adamantane carboxylic acid (Acros, 99 %)
and 1.04 g 1,2-hexadecanediol (HDD, Sigma Aldrich). The reaction mixture was put under vacuum
at 80 °C for 30 minutes and then heated at a rate of 10 °C min-1. When the temperature reached 170
°C, 334 mg Co2(CO)8 (Acros, 95 %) , dissolved in 3.2 mL 1,2-dichlorobenzene, were injected. The
reaction mixture was further heated to 230 °C for 40 minutes. Then, the reaction mixture was
allowed to cool down and 50 mL toluene were injected at 200 °C. When the system was cooled
down to 90 °C, the reaction mixture was divided into 6 centrifuge tubes (50 mL) and 30 mL of warm
isopropanol (~50 C°) were added to each centrifuge tube. After 2 minutes of centrifugation at 6000
rpm, the supernatant was removed and the precipitate was dispersible in non-polar solvent such as
hexane and toluene. After washing the colloid with isopropanol 3 times, the NCs were kept in
hexane.
For the synthesis of Pt3Co2 NCs, the method employed for Pt3Co NCs synthesis was
modified. Hexadecylamine was replaced by OAm (41 mL), diphenyl ether was replaced by 1-
octadecene (Acros, 90 %,16 mL), and HDD was not added. After evacuation at 80 °C for 30 minutes,
the reaction mixture was heated to 300 °C. At 170 °C, the solution of Co2(CO)8/1,2-dichlorobenzene
was injected in the same amount as the Pt3Co synthesis. After 30 minutes at 300 °C, the reaction
mixture was cooled down to room temperature and 50 mL hexane were added. The work up was
the same as for the synthesis of Pt3Co NCs. Notably, in the case of Pt3Co, HDD was used to reduce
(201). On the other hand, for the synthesis of Pt3Co2,
hexadecylamine was replaced by oleylamine and HDD was not added. As oleylamine is a relatively
milder reducing agent than HDD (202), this reaction resulted in NCs relatively richer in cobalt
compared to the reaction done by the polyol process.
From the TEM images of NCs in Figure 4.7a-c, the average diameters of NCs were 2.4 nm,
3.2 nm, and 3.7 nm with less than 8% of size distribution for Pt, Pt3Co, and Pt3Co2 NCs, respectively.
A superlattice structure for each composition of NCs was observed, indicating that the NCs were
highly monodisperse. Wide angle x-ray scattering data show that all the NCs possess face-centered
cubic (fcc) crystal structure as displayed in Figure 4.7d. The compositions of the NCs were
confirmed by Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES).
127
Figure 4.7
TEM images of a) Pt NCs, b) Pt3Co NCs, and c) Pt3Co2 NCs, after removing the
solvent. The scale bars in the insets represent 2 nm. d) Wide angle X-ray
scattering data of Pt (red), Pt3Co (green), and Pt3Co2 (blue).
To prepare the final, 10 wt% carbon supported NC catalyst, the NCs suspended in 20 mL
hexane were mixed with carbon powder (Cabot, Vulcan XC72R) in a 50 mL centrifuge tube under
sonication. After 15 minutes of sonication, the solution was centrifuged at 6000 rpm for 1 minute.
The transparent supernatant was removed and the powder was washed with 20 mL isopropanol
and centrifuged again. Then, the powder was dried in vacuum at 50 °C.
The organic ligands covering the NCs surface for colloidal stability have the drawback that
they inhibit catalytic reactions due to limited access of reactants to the binding sites on the NC.
Therefore, the surface ligands should be removed in order to maximize NCs available surface area.
The surface cleaning procedure consists of two steps. First, 300 mg of 10 wt% NCs/C catalysts were
exposed to O2 plasma for 15 min. After the plasma treatment, the catalyst was put into a furnace at
500 °C and taken out after 1 minute (Rapid Thermal Annealing ─ RTA) (40). Figure 4.8 shows TEM
images of as-deposited and surface treated NCs on carbon support. After the plasma and thermal
128
treatment, no significant change in size and morphology is observed in all cases, which proves the
reliability of the surface treatment technique.
Figure 4.8 TEM images of (A) as-synthesized NC catalyst on carbon support and (B)
after surface treatment.
Also, from wide angle x-ray scattering measurements (Figure 4.9), it is confirmed that the
phase of Pt3Co and Pt3Co2 is not transformed from disordered fcc to ordered fct structure.
Therefore, the potential effects of crystal structure ordering on the catalytic activity can be
excluded and any difference in activity can be attributed to compositional effects (203).
Figure 4.9 Wide angle x-ray scattering data of as-synthesized (red) and surface cleaned
(blue) Pt3Co2 NCs.
129
4.3.3 HDO studies
Previous studies on carbon-supported Pt and Co catalysts have shown that
hydrodeoxygenation of HMF to DMF is a series reaction (185), proceeding as indicated in Scheme
4.2. The HMF (A) first reacts to a group of partially hydrogenated intermediate compounds (B),
including 2-propoxymethyl-5-furanmethanol (ether-furfuryl alcohol, or EFA), 2-propoxymethyl-5-
methylfuran (ether-methyl furan, or EMF), 5-methyl furfural (MFu), 2-hydroxylmethyl-5-methyl
furan (HMMF), 2,5-bis(hydroxymethyl)furan (BHMF), and 2,5-bis(propoxymethyl)furan (BEF). These
intermediate compounds can all be converted to DMF (C), which in turn reacts to over-
hydrogenated products (D), dimethyltetrahydrofuran (DMTHF), 2-hexanone, 2-hexanol, 2,5-
hexanedione, and their etherification derivatives, 1-propoxy-1-methyl-pentane (2-propoxyhexane)
and 1,4-dipropoxy-1,4-dimethyl-butane (2,5-dipropoxyhexane).
Scheme 4.2 Reaction network for liquid-phase hydrodeoxygenation of HMF in alcohol
solvent.
Figure 4.10 reports the conversion and products yield as a function of space time on the
NCs-based catalyst prepared in this study. 10 wt% Pt NCs/C catalysts (Figure 4.10a) were much
active: even for the shortest space time and at low temperature (120°C), the HMF conversion was
greater than 65%. Scheme 4.2 also applies to the reaction of HMF in this case. Initially, partially
hydrogenated products, B, were formed in the highest yields, but these declined steadily with
increasing space time. DMF yield initially increased, then decreased, providing strong evidence
that DMF is an intermediate product in a series reaction. The maximum yield was approximately
41%. Over-hydrogenated compounds, D, only formed at higher space times, indicating that they
are not primary products. The fact that their formation follows the consumption of DMF strongly
130
suggests they are formed from DMF. The HDO of HMF over a traditionally prepared Co/C catalyst
has been reported in a previous study(185). The products formed as a function of space time were
similar to that formed over Pt catalysts. Co itself is not selective to DMF, due to the formation of the
over-hydrogenated product 2,5-hexanedione at high space times.
Figure 4.10
Conversion and product distribution for the HDO reaction of HMF over (a)
10 wt% Pt NCs/C, (b) 10 wt% Pt3Co NCs/C, (c) 10 wt% Pt3Co2 NCs/C, as a
function of reactor space time. Reaction conditions: 33 bar and 120 ºC. ()
HMF conversion, () product group B, () DMF, () product group D.
Similar experiments were performed on 10 wt% Pt3Co NCs/C and 10 wt% Pt3Co2 NCs/C
catalysts. Data at 120°C and 33 bar are shown in Figure 4.10b and 4.10c. For a given space time, the
HMF conversions over the Pt3Co in Figure 10.4b were slightly lower than that obtained on the pure
Pt NCs. However, the initial products were the same partially hydrogenated compounds, B (see
Table 4.2), with these again being converted to DMF at a similar rate. However, on the Pt3Co
catalyst, the DMF yield continued to increase, to a value of 75%; and only relatively small quantities
of over-hydrogenated compounds, D, were formed at the largest space times. The activity of the
Pt3Co2 catalyst was noticeably lower than that of the other two samples, and 100% conversion of
HMF was achieved only at the highest space time. Because of the lower activity, the DMF yield was
131
still increasing at the highest space time and the production of D-group compounds was
negligible.
A detailed analysis of the partially hydrogenated (B) and over-hydrogenated (D)
compounds formed during HDO of HMF over the NCs-based catalysts is given in Table 4.2 and
Figure 4.11.
Table 4.2 Yields of partially hydrogenated compounds (group B) for HDO of HMF at
120 °C and 33 bar.
132
Figure 4.11
Yields of over-hydrogenated compounds (group D) as a function of space
time at 33 bar. At 120°C: (a) 10 wt% Pt NCs/C, (b) 10 wt% Pt3Co NCs/C and
(c) 10 wt% Pt3Co2 NCs/C. At 160°C: (d) 10 wt% Pt3Co NCs/C and (e) 10 wt%
Pt3Co2 NCs/C. (□) DMTHF, (▽) 2-hexanone, 2-hexanol and 2-
propoxyhexane, (△) 2,5-hexandione, 2,5-dipropoxyhexane, (◯) hexane, (◁)
unidentified.
133
Due to the lower rates observed for the Pt-Co catalysts, additional reaction measurements
were performed at 160°C and 33 bar in order determine the evolution of products, with results
shown in Figure 4.12. As shown in Figure 4.12a, the HMF conversion was nearly 90% on the Pt3Co
NCs-based catalyst, even at the lowest space time. The B-products again declined steadily with
increasing space time but DMF yield went through a maximum of about 75% at this temperature,
with over-hydrogenated products being produced from DMF. On the other hand, DMF yields up to
98% were obtained over the Pt3Co2 NCs-based catalyst (Figure 4.12b) for the highest space time.
Figure 4.12
Conversion and product distribution for the HDO reaction of HMF over (a)
10 wt% Pt3Co NCs/C, (b) 10 wt% Pt3Co2 NCs/C, as a function of reactor space
time. Reaction conditions: 33 bar and 160 ºC.
() HMF conversion, () product group B, () DMF, () product group D.
The Pt3Co2 sample was also very stable compared to the Pt catalyst. Figure 4.13 shows the
HMF conversion and DMF yield for the two catalysts as a function of time at 160°C and a space of
1.0 g min mL-1. The Pt3Co2-based catalyst has no observable deactivation or change in selectivity for
a period of at least 14 h. By contrast, the Pt-based catalyst deactivated rapidly under the same
conditions. It should be noted that Pt/C catalyst was highly active under these conditions, so that
the low initial yield is due to the reaction of DMF to over-hydrogenated (D) compounds. The
increasing yield with short times results from the lower catalyst activity.
134
Figure 4.13
Time on stream measurements for HMF hydrodeoxygenation. Reaction
conditions: 33 bar, 160 ºC, W/F 1.0 g min mL-1. () HMF conversion over 10
wt% Pt3Co2 NCs/C, () DMF yield over 10 wt% Pt3Co2 NCs/C, () HMF
conversion over 10 wt% Pt NCs/C, () DMF yield over 10 wt% Pt NCs/C.
In addition to the stability against coking, the alloy catalyst is also more thermally
stable(204). Statistical particle size analysis was performed on TEM images for Pt and PtCo catalysts
after 5 hours reaction at 160 °C (Figure 4.14). Pt NCs underwent morphological changes, with the
average particle size increasing from 2.4 nm to 3.5 nm and a significant size distribution change
from 7.3% to 19%. By contrast, Pt3Co2 NCs showed almost no morphological change. The average
size of the particles were unaffected and the size distribution increased only slightly from 7.3% to
10%.
135
Figure 4.14 The particle size distribution of (a) Pt and (b) Pt3Co2 nanocrystals before and
after 5 h of HDO reaction conditions at 160°C.
The results in Figure 4.10 and Figure 4.12 indicate that Co alloying with Pt has a
modest effect on HDO rates for HMF but strongly suppresses reactions of DMF. To investigate this
in more detail, we examined the reaction of DMF on the same three catalysts, with conversions and
product distributions shown in Figure 4.15. The reactions were carried out at 33 bar and either 120
°C (for Pt) or 160°C (for the Pt-Co alloys), using 1-propanol solutions with the same DMF molar
concentration as that used in the HMF experiments. Since water is formed in the reaction of HMF,
these experiments do not perfectly mimic the sequential reaction found in HMF; however, some
water was also formed in the DMF reaction measurements by dehydration of alcohols to form
dipropyl ether.
136
Figure 4.15
Conversion and product distribution for the reaction of DMF as a function of
space time at 33 bar: (a) 10 wt% Pt NCs/C at 120 ºC; (b) 10 wt% Pt3Co NCs/C
at 160 ºC, (c) 10 wt% Pt3Co2 NCs/C at 160 ºC.
() DMF conversion, () DMTHF, () 2-hexanone, 2-hexanol and 2-
propoxyhexane, () 2,5-hexandione, 2,5-dipropoxyhexane, () hexane.
As shown in Figure 4.15a, DMF is converted rapidly on Pt/C, even at 120°C. The main
products are the open-ring ketones and ethers, which were also formed at high space times for the
reaction of HMF (see Figure 4.11). The reaction of DMF on the Pt-Co catalysts was carried out at
160°C because of their lower activities. Even at this higher temperature, the DMF conversion on the
Pt3Co sample was lower than that observed on the Pt catalyst, although still significant. The
products on the Pt3Co catalyst were essentially the same as the ones observed on Pt. However, the
conversion of DMF on the Pt3Co2 sample at 160°C, shown in Figure 4.15c, was very low for all space
times, reaching a value of only 10% at a space time of 1.0 g min mL-1.
137
4.3.3 Active Site Structure Investigation
In order to understand the role of Co for high HMF-to-DMF selectivity and elucidate the
nature of the active site, a combination of microscopic, spectroscopic, and computational tools
were employed, as discussed below. First of all, wide angle X-ray scattering data (Figure 4.7d) show
that the (220) peak shifts from 67.6° for Pt NCs to 68.2° and 68.4° for Pt3Co and Pt3Co2, indicating
the replacement of Pt by Co in the lattice structure. The lattice constants of the NCs, determined
from the position of the (220) peak on the x-ray scattering patterns, are 3.92, 3.87, and 3.87 Å for Pt,
Pt3Co, and Pt3Co2 NCs, respectively. Usin 3Co and Pt3Co2
NCs was estimated to be 13.4 mol.%. The fact that the bulk Co fraction is lower than that of the
alloy stoichiometry is an initial indication of Co segregation.
The local environments of the Pt and Co atoms were further investigated using X-Ray
Absorption Spectroscopy (XAS). The Pt L3 near-edge spectra of Pt3Co2 samples reduced in H2 at 250
and 400oC are shown in Figure 4.16a, together with the Co K near-edge spectra (Figure 4.16b). On
the Pt edge, the white line and edge positions of the samples coincide with those of the Pt foil for
both reduction temperatures, demonstrating that the Pt is fully reduced in all cases. However, the
Extended X-Ray Absorption Fine Structure (EXAFS) on the Pt edge suggests that there is surface
segregation of Co in the Pt-Co alloy particles, as the Pt:Co ratio in the Pt coordination sphere (3.1 ±
0.6) is greater than the nominal. Considering the fact that the cobalt precursor was injected at 170
°C in the synthesis, it is reasonable that the Pt-Co alloy NCs had cobalt rich shell, because platinum
precursor could decompose and nucleate at lower temperature, forming the core of NCs. The
fitting parameters for these spectra are reported in Table 4.3. The Co edge EXAFS was not fitted for
the sample reduced at 250 °C, as the variables required to completely describe the structure are
greater than the number of available independent data points. After reduction at 400 °C, the Pt:Co
ratio in the Pt coordination sphere is closer to the nominal (2.1 ± 0.4), consistent with at least partial
reverse Co segregation to the bulk. A similar reverse segregation phenomenon has been reported
previously for PtNi nanoparticles (205).
138
Figure 4.16
(a) Pt L3 edge XANES. Pt foil (red), Pt3Co2 reduced at 250 ºC (blue) and Pt3Co2
reduced at 400 ºC (black). (b) Co K edge XANES. Co foil standard (black), CoO
standard (red), Pt3Co2 reduced at 250 ºC (yellow) and Pt3Co2 reduced at 400
ºC (green).
Table 4.3 Fitting of the EXAFS Pt and Co spectra. All distances are in Angstroms.
139
The X-Ray Absorption Near Edge Spectra (XANES) of the Co K edge, reported in Figure
4.16b, provides further information on the nature of the alloy NCs. First, the spectra indicate that
Co remains partially oxidized, even after reduction at 400°C. Using a linear combination of spectra
from CoO and Co standards to fit the results for the alloy catalyst, the average Co oxidation states
after reduction at 250°C and 400 °C were 1.2 (60% CoO) and 0.72 (36% CoO), respectively. Surface
oxygen is known to induce Co surface segregation in a Pt-Co alloy (206). In turn, Co forms a surface
monolayer oxide with properties distinct from those of the bulk CoO (207). As step sites constitute
ca. 30% of all surface sites for 3-4 nm nanoparticles (208) and tend to stay more oxidized than
terraces (209), 30% of surface sites were assumed to be composed of Co in the +2 oxidation state
(210). Following an analogous Fe3O2/Pt(111) structure observed using STM (211), the rest of the
monolayer surface oxide was assumed to be Co3O2 with a honeycomb structure on a Pt-Co metallic
core as a nanoparticle model (Figure 4.17a).
Figure 4.17
(a) Pt3Co2 nanocrystal model involving an alloy core (88% Pt, 12% Co based
on XRD) covered with a Co3O2 surface oxide monolayer with a honeycomb
structure; (b) and (c) correspond to Pt and Pt3Co NC models.
Table 4.4 compares XAS/XRD results with geometric estimates of the alloy core/oxide shell
spherical nanoparticle model. Overall, the model exhibits excellent agreement with the
experimental data, given the approximations invoked. An average Co oxidation state of 1.18 was
140
predicted, close to the experimental estimated value of 1.20. The low Co oxidation state is
consistent with an O:Co atomic ratio of less than 1 on the majority of surface sites. In contrast, a
previously observed surface oxide with CoO stoichiometry (207) would yield an average oxidation
state >1.5. The predicted Co content in the bulk alloy (14.4 mol w
estimation from XRD (13.4 mol%). The experimental and geometric Pt-Co and Pt-Pt CNs agree well.
The XAS data demonstrate that Pt3Co2 NCs, reduced at 250°C, consist of a Pt-rich core with the
majority of Co segregated to the surface, forming a CoOx surface oxide shell. The Co3O2 honeycomb
monolayer as a dominant surface structure is consistent with XAS results.
Table 4.4
Average coordination numbers and Co oxidation states for Pt3Co2 NCs
reduced at 250oC, determined by X-ray absorption spectroscopy and
estimated using a spherical core/shell NC model with planes covered by a
Co3O2 surface oxide monolayer (70%) and step sites covered by CoO (30%).
Further evidence for Co3O2 surface oxide formation comes from CO chemisorption
-Al2O3, shown in Table 4.5.
Table 4.5 Metal dispersions of alumina supported catalysts after 250 °C and 400 °C
reduction, assuming CO/Pt=1 and no CO adsorption on Co atoms.
141
CO chemisorption experiments were performed on alumina supported NCs in order to
avoid sorption problems related to active carbon supports (212). Before the chemisorption
experiment, the samples were evacuated at 400oC and reduced for 30 minutes at 250oC or 400oC in
200 Torr of H2. In determining the NCs dispersion, an adsorption stoichiometry of one CO molecule
per surface Pt was assumed(213). Chemisorption of CO on Co was not included in the calculation of
the dispersion because control experiments on Co/Al2O3 catalysts prepared by impregnation
method did not show any CO adsorption. After reduction at 250°C, CO adsorption on the Pt3Co2
NCs is negligible, consistently with the Pt atoms being covered. After 400°C reduction, CO
chemisorption is comparable to what was observed with Pt NCs, due to reverse segregation of Co
to the bulk. DFT results indicate that CO interacts weakly with Co3O2/Pt(111) compared to Pt(111)
(ca -0.7 vs. ~-2 eV binding energy(211), respectively), consistent with the lack of CO adsorption. A
similar in magnitude CO binding energy was correlated with no CO adsorption in XPS
measurements on Fe3O2/Pt(111) honeycomb structure (211). The importance of a Co3O2 overlayer
structure is further supported from DFT calculations, discussed next.
4.3.4 Theoretical Insights into the Reaction Mechanism, Catalyst Composition Effects, and Stability
DFT calculations9 were performed in order to understand the mechanism of the HDO
reaction, catalyst stability issues, and the differences among the three catalysts (Pt, Pt3Co, and
Pt3Co2). Regarding the reaction mechanism on the Pt3Co2 catalyst, the calculations showed that the
Co3O2 honeycomb monolayer supported on the Pt rich core is capable of catalyzing key reaction
steps involved in the HMF to DMF conversion. The overall reaction is assumed to proceed via the
following steps: 1) H2 dissociation, 2) C=O hydrogenation, and 3) selective HDO with concomitant
oxygen removal from the surface in the form of water. The H2 dissociation step can occur via
several homolytic and heterolytic dissociation paths (210), but the calculations indicate that the
homolytic splitting of a weakly physisorbed H2 molecule (-0.1 eV binding energy) over a single Co
atom is most energetically favorable (a 0.7 eV reaction barrier). The final state (0.3 eV more stable
than gaseous H2) entails both H atoms bound to Co and Pt atoms in bridging configurations.
9 Calculations have been carried out under the generalized gradient approximation using VASP
software (288 290). Kohn-Sham eigenstates have been expanded in a plane wave basis set with a kinetic energy cutoff of 400 eV. Sampling of the first Brillouin zone has been carried out according to the Monkhurst-Pack (291) 3x3x1 k-point mesh. The initial magnetic moment of Co atoms has been set to 2.0 Bohr-magnetons. Exchange, correlation, and dispersion effects have been approximated with a PBE-D3 (292, 293) functional. Threshold value for maximum atomic forces has been set to 0.05 eV Å-1. Transition states have been identified via a climbing-image nudged elastic band (CINEB) and/or a dimer (294 297) method with the
-1. A honeycomb Co3O2 on a Pt-rich metallic core has been modeled as a Co3O2/Pt(111) surface using a (4x4) supercell with a honeycomb Co3O2 structure placed on top of three Pt layers (bottom two were fixed). A finite difference method was used to calculate selected vibrational frequencies. DFT energies in a vacuum are reported, unless otherwise stated.
142
Prior to the C=O hydrogenation step, HMF weakly adsorbs on the surface (-0.8 eV BE);
hydrogenation of the C=O carbonyl group exhibits a low reaction barrier when it is weakly bound
to the surface (214, 215). Specifically, a concerted addition of two H atoms to HMF occurs with a 0.8
eV barrier, yielding BHMF. The overall hydrogenation is exothermic (reaction energy of -0.6 eV).
BHMF can either desorb (0.9 eV desorption energy) or undergo HDO, ultimately forming DMF.
The subsequent HDO mechanism of BHMF on the Pt-rich particle with Co3O2 coating is
radical in nature, and is depicted in Figure 4.18. A similar mechanism was proposed in the HDO of
furfural to form 2-methylfuran on a Ru/RuO2 catalyst, consistently with a range of experimental and
computational data (210). BHMF undergoes C-O bond scission on a honeycomb edge site
consisting of two Co atoms, forming a loosely bound radical and an OH group, with a reaction
energy of +0.9 eV and a barrier of 1.2 eV (Figure 4.18, steps 1-2). A hydrogen atom transfers from
the OH to the radical, yielding HMMF and a chemisorbed oxygen atom (Figure 4.18, step 3-4). C-O
scission occurs similarly on the second hydroxymethyl group (not shown), forming DMF as the final
product. The chemisorbed O atom (+1.7 eV binding energy with respect to H2 and H2O) reacts
rapidly with H2 (a -1.2 eV exothermic dissociative adsorption with a 0.3 eV barrier) to form co-
adsorbed OH and H that subsequently recombine with a 0.2 eV barrier (-0.9 eV reaction energy) to
form water(Figure 4.18, steps 5-6). Finally, water desorbs with a +0.4 eV energy to complete the
catalytic cycle.
HMF, BHMF, HMMF, and DMF weakly interact with the Co3O2 surface (~ -0.8 eV BE,
dominated by dispersion forces (216), as opposed to chemisorption on metal atoms with BE of the
order of -2 eV(217)), largely retaining a gaseous-like molecular geometry(210). The absence of
covalent bonding of the ring with the metal surface is key to rationalizing the high selectivity of the
catalyst, because opening of the furan ring and decarbonylation require strong chemisorption of
the ring in a flat geometry, with partial sp2→ sp3 re-hybridization of ring carbon atoms (184, 218
222). Lack of covalent bonding between the ring with the Co3O2 oxide protects the ring from
further side reactions and explains the low reactivity of DMF. The Co3O2 surface layer is capable of
catalyzing C-O bond hydrogenolysis in HMF that leads to selective production of DMF.
143
Figure 4.18
Reaction mechanism of BHMF hydrodeoxygenation to HMMF on the
Co3O2/Pt(111) surface. DFT reaction barriers (energies) are given in eV. The
inset depicts a portion of a Co3O2/Pt(111) surface. Two Co atoms
participating in C-O bond activation are encircled with a white ellipsoid.
In order to assess the catalyst stability at a H2-rich environment, the rate of initiation of
Co3O2 reduction via vacancy formation was calculated (210). Under experimental HDO conditions
(160oC, 33 bar H2), the vacancy formation rate is a factor of 2 lower than under in situ XAS
conditions (250oC, 1 bar H2). Furthermore, the vacancy, once formed, is easily oxidized by BHMF-to-
HMMF reactions. This analysis provides a rationalization as to why the Co3O2 surface oxide is stable
in a reducing reaction environment.
In contrast to the highly selective, oxide-covered Pt3Co2 catalyst, Pt carries out facile
hydrogenation of the C=O group to BHMF, but dehydroxylates BHMF to form DMF slowly (210).
Furthermore, the DMF ring interacts strongly with Pt, promoting ring hydrogenation and ring
opening with barriers which are lower than that of the dehydroxylation reaction (210). The
144
computed barriers for HDO of HMF are comparable to the barriers for ring hydrogenation and ring
opening of DMF, consistent with the observation that selectivity to DMF is modest.
Pt3Co NCs exhibit catalytic properties intermediate between Pt and Pt3Co2. Unlike the
Pt3Co2 NCs, there are not enough Co atoms to completely cover the surface with an oxide
monolayer. In this catalyst, the surface is predicted to consist of 1/2 Co oxide and ~1/2 Pt atoms.
We propose that this difference in structure exposes Pt patches on the surface (Figure 4.17b). The
presence of Pt sites at the surface leads to the partial non-selectivity of the Pt3Co catalyst.
4.3.5 Discussion
The development of better catalysts for HDO of HMF requires an understanding of the
reaction mechanism. First, it is important to recognize that the reaction is sequential (41, 185). The
poor selectivity that is observed with many metals is due to the fact that they further catalyze
reactions of DMF, the desired product. While it is required that a catalyst has good activity for HDO
of HMF, a selective catalyst must also be a poor catalyst for reactions of DMF. The sequential nature
of the reaction also makes it essential that no part of the catalyst is nonselective. For reactions in
which both the desired and side products form in parallel, having a small percentage of the catalyst
surface showing a lower selectivity will not dramatically change the overall selectivity. With a
sequential reaction, the nonselective part of the catalyst can have a much more dramatic effect.
This has important consequences for alloy catalysts. While the catalyst based on Pt3Co2 NCs has the
necessary properties to achieve a very high selectivity, alloy catalysts prepared by conventional
impregnation methods will not be so compositionally uniform. Both Pt and Co are individually
nonselective because they are active for reactions of DMF, so that any metal in the catalyst which is
not in the form of an alloy will be nonselective. Indeed, a Pt3Co2 catalyst prepared by incipient
wetness of the metal salts showed a maximum selectivity of less than 80%(210). In this context, it is
interesting to consider the work from Schüth and co-workers(188), who first reported extremely
high selectivity for HDO of HMF with PtCo alloys. In their case, the highest selectivity were achieved
when the metal particles were encapsulated in porous carbon spheres. We suggest that those
carbon spheres were important for achieving a high compositional uniformity in the particles.
Control of the metal composition is essential in order to achieve good HDO selectivity.
From the HDO studies, we learned that Pt3Co was not as selective as Pt3Co2. The reason for this is
the difference in particle surface structure. Pt3Co does not have enough Co atoms to completely
cover the surface. The incomplete coverage of Co oxide leads to the presence of uncovered Pt sites,
and therefore causes partial non-selectivity of the Pt3Co catalyst. While it may be possible to
increase the Co:Pt ratio further, a catalyst with excess of Co is likely to be unselective since
monometallic Co catalysts are unselective (185).
145
Catalyst stability is equally important to activity and the present results suggest that there
is a direct correlation between stability and selectivity. The most serious and rapid deactivation in
the experiments was due to coking which is caused by further reaction of over-hydrogenated
products, such as 2,5-hexanedione. The Pt-Co alloy catalyst also seems to be more tolerant against
sintering, possibly as a result of the core-shell structure.
4.3.6 Conclusions
High selectivity of DMF from liquid-phase HDO of HMF with H2 was achieved over a well-
controlled Pt-Co /C catalyst. Particularly, over Pt3Co2 nanocrystal-based catalyst, 98% of DMF yield
was obtained with the optimized reaction temperature and space time. Recognizing the sequential
nature of the HMF HDO reaction is the key for catalyst-development strategies. Noble metals
interact strongly with the furan ring to promote side reactions. The fundamental principle for the
superior performance of Pt3Co2 is that the bimetallic alloy forms a monolayer oxide on the surface
of the metallic core that interacts weakly with the furan ring to prevent over-hydrogenation and
ring opening of DMF to secondary by-products while forming active sites to carry out the HDO
process. In this regard, composition control is crucial to cover the entire surface with an oxide layer
and avoid exposed metallic patches that could promote side reactions. In the following sections,
other metallic alloys will be studied in the HDO of HMF, in order to get further insights in the
factors controlling selectivity to DMF.
146
4.4 Base Metal-Pt Alloys for HDO of HMF
4.4.1 Introduction
There is considerable evidence that bimetallic catalysts can be more selective in HDO
reactions than their monometallic counterparts. For example, in the gas-phase reaction of furfural,
Ni-Fe catalysts were found to exhibit good selectivity to methylfuran under conditions which
resulted in the decarbonylation product, furan, over monometallic Ni catalysts (38). As discussed in
the previous section, Pt-Co nanocrystals (NCs) exhibited much better selectivity and stability than
the monometallic catalysts in the HDO reaction of HMF(210). The high selectivity of the Pt-Co
catalysts resulted from the fact that DMF is not reactive over the bimetallic catalyst.
It is not entirely clear why the bimetallic catalysts are more selective for HDO and the
reasons may vary with the particular alloy or reactor conditions. Several studies have argued that
high selectivity is due to the oxophilicity of the secondary metal (38, 223 225). For example,
Resasco and coworkers (38) observed that NiFe catalysts showed superior selectivity compared to
Ni for the reaction of furfural to methyl furan. Their theoretical calculations indicated that the
2(C,O) configuration of the aldehyde carbonyl, weakening the
C-O bond and suppressing the formation of acyl intermediates that undergo decarbonylation (38).
Similarly, for Pt-Co bimetallic catalysts studied in the previous section, Co tends to form a CoOx
shell on a Pt-rich core and this oxide layer prevents interactions between the Pt and the furan ring,
enhancing DMF yields from HDO of HMF (210).
However, high selectivity has also been reported for HDO reactions over Pd-Cu alloys, for
which an oxide overlayer is not expected (226). Although Cu is not oxophilic, DFT calculations
suggested that Cu atoms at the surface repel the aromatic furan ring (227), causing furfural to
1(O)-aldehyde configuration via the carbonyl oxygen. Such an
orientation of the furan ring prevents side reactions on the ring itself (38, 226, 227).
In the present section, the catalytic properties of carbon-supported nanocrystalline Pt-Ni,
Pt-Zn, and Pt-Cu alloys will be studied, to determine whether Pt-Co is unique among the Pt alloys in
providing high selectivity for HDO of HMF. Solvothermal methods (see Section 4.3.1) were
employed in order to produce NCs having homogeneous and controlled shape, dimension and
composition and to investigate the optimal compositions for selective alloy catalysts. While Ni is
oxophilic and catalytically similar to Co, Zn and Cu have properties that contrast sharply with that
of Co and provide a test for the properties required for high selectivity to DMF (>95%). Surprisingly,
each of the alloys exhibited superior selectivity compared to the monometallic catalysts.
147
4.4.2 Catalysts Synthesis and Characterization
The following chemicals were used in the synthesis of the investigated NCs: nickel (II)
acetate tetrahydrate (Ni(ac)2·4H2O, Sigma-Aldrich, 98 %), platinum (II) acetylacetonate (Pt(acac)2,
Acros, 98%), nickel (II) acetylacetonate (Ni(acac)2, Acros Organics, 96 %), copper (II) acetylacetonate
(Cu(acac)2, Sigma-Aldrich, zinc (II) acetylacetonate (Zn(acac)2, Acros Organics, 25 % Zn),
1,2-hexadecanediol (HDD, Sigma-Aldrich, 90 %), trioctylamine (TOA, Acros Organics, 97 %),
oleylamine (OAm, Sigma-Aldrich, 70 %), oleic acid (OAc, Sigma-Aldrich, 90 %), and
trioctylphosphine (TOP, Sigma-Aldrich, 97 %), borane, tert-butylamine complex, (BTB, Strem
Chemicals, 97 %), diphenyl ether (DPE, Sigma-Aldrich, 99 %) and 1,2-dichlorobenzene (DCB, Acros
Organics, 99 %). Nickel(II) nitrate hexahydrate (Ni(NO3)2 6H2O, Alfa Aesar, 98%), Zinc(II) nitrate
hexahydrate (Zn(NO3)2 6H2O, Alfa Aesar, 99%), Copper(II) nitrate trihydrate (Cu(NO3)2 3H2O, Alfa
Aesar, 98%), and Cobalt(II) nitrate hexahydrate (Co(NO3)2 6H2O, Aldrich, 99%).
Pt6Ni and PtNi NCs were synthesized using a scaled-up method reported previously (228).
For the Pt6Ni NCs, Ni(ac)2·4H2O (0.33 mmol) and HDD (1.15 mmol) were dissolved in a solution of
DPE (40 mL), OAm (0.8 mL), and OAc (0.8 mL). The reaction mixture was kept at 80 °C for 30 min
under vacuum and then heated to 200 °C under a nitrogen atmosphere. Pt(acac)2 (0.67 mmol),
dissolved in DCB (2.4 mL), was injected into this mixture at 200 °C. The resulting solution was kept
at 200 °C for 1 h before cooling to room temperature. The resulting NCs were then purified by
precipitation with ethanol and centrifugation at 8000 rpm for 5 min. The precipitate was washed
twice with hexane/ethanol (1:3) mixtures before the final NCs were dispersed in hexane. The same
procedure was used to synthesize PtNi NCs, except that the amounts of Ni(ac)2·4H2O (2 mmol) and
OAc (1.0 mL) were adjusted. Also, after the washing steps, the PtNi NCs were re-dispersed in
hexane and size-selective precipitation was performed (229).
To synthesize Pt3Ni NCs, Ni(acac)2 (0.4 mmol) and Pt(acac)2 (0.4 mmol) were dissolved in a
solution of TOA (40 mL), OAm (5.44 mL), OAc (1.28 mL), and TOP (0.45 mL). The reaction mixture
was kept under vacuum at 70 °C for 30 min, then heated to 330 °C under a N2 atmosphere. After 30
min, the reaction mixture was cooled to room temperature and purified by precipitation with a
mixture of isopropanol and ethanol. After being centrifuged at 8000 rpm for 2 min, the precipitate
was washed twice with hexane/ethanol (1:3) mixtures. Finally, the NCs were dispersed in hexane.
PtCu and Pt2Zn NCs were synthesized by modifying a previously reported method (202). For
PtCu NCs, Pt(acac)2 (0.4 mmol), Cu(acac)2 (0.4 mmol), and BTB (1.1 mmol) were dissolved in OAm
(20 mL). The reaction mixture was kept under vacuum at 80 °C for 30 min and then heated to 300
°C at a rate of 5 °C min 1. After 1 h, the reaction mixture was cooled to room temperature.
Purification of the NCs was achieved by addition of isopropanol, followed by centrifugation at 8000
148
rpm for 2 min. The precipitate was washed three times with hexane/ethanol (1:3) mixtures and the
final NCs were dispersed in hexane. For Pt2Zn NCs, Pt(acac)2 (0.5 mmol), Zn(acac)2 (0.5 mmol) and
BTB (1.1 mmol) were dissolved in OAm (20 mL). The reaction conditions and purification steps were
the same as those for PtCu NCs synthesis.
To prepare the carbon-supported catalysts from the NCs, the NCs, dispersed in hexane,
were mixed with carbon powder (Cabot, Vulcan XC72R) to a loading of 10 wt% metal. After
sonication for 15 min, the solution was centrifuged at 6000 rpm for 1 min. The supernatant was
removed and the precipitate was washed twice with isopropanol, followed by centrifugation. After
drying each sample in a vacuum oven overnight at 50 °C, the samples were first treated with an O2
plasma cleaner (18 W, Harrick Plasma) for 15 min, then transferred for 1 min into a muffle furnace
that had been preheated to 500 °C (40), as discussed in Section 4.3.1.
Reference 10 wt% Ni/C, Co/C, Cu/C and Zn/C catalyst were prepared by impregnation,
using a previously reported procedure (185). The Ni, Zn, Cu and Co precursors were Ni(NO3)2 6H2O,
Zn(NO3)2 6H2O, Cu(NO3)2 3H2O, and Co(NO3)2 6H2O, respectively. Each metal precursor was firstly
dissolved in a water/ethanol (5:1) solution and then mixed with carbon black (Vulcan XC-72R). After
drying the catalyst at room temperature, the resulting powder was heated in flowing He to 500 °C
for 6 h, using a heating ramp of 3 °C min 1. The dried powders were reduced by flowing a 5% H2/He
mixture over the catalysts at 60 mLmin 1 while ramping the temperature at 2 °C min 1 to 400 °C,
followed by heating to 500 °C with heating ramp of 1 °C min 1. The catalysts were then held at this
temperature for 2 h.
Nearly monodisperse, Pt-based alloyed NCs with various compositions were synthesized by
solvothermal synthesis, which enabled us to investigate the composition dependent catalytic
behavior of the bimetallic NCs. TEM images of the Pt6Ni, Pt3Ni and PtNi NCs are shown in Figure
4.19a-c and indicate that, in each case, the particles are uniform and spherical in shape. Based on
the small-angle X-ray scattering (SAXS) data fitted to Rayleigh function (Figure 4.19d), the average
sizes and size distributions of the Pt6Ni, Pt3Ni, and PtNi NCs were 3.0±0.4, 3.0±0.6 and 6.0±0.8 nm,
respectively. As shown in Figure 4.19e, the XRD patterns for each of the NCs showed a face-
centered cubic (fcc) crystal structure, and shifts to higher angles in the peaks at 2 were observed
for Pt6Ni and PtNi NCs, compared with the pure Pt NCs. This results from a contraction of the lattice
based on the Vegar (230) and is a good indication that the NCs are well-mixed bimetallic
alloys. In the case of Pt3Ni NCs, the only observed 2 peak corresponds to the (111) plane. The
absence of additional peaks suggests either polycrystallinity or the presence of a high
concentration of defects, as is often seen with NC systems (231).
149
Figure 4.19
The TEM images of (a) Pt6Ni, (b) Pt3Ni and (c) PtNi NCs, the corresponding (d)
SAXS patterns and (e) XRD patterns. Inset in (a) shows the HRTEM image of
Pt6Ni NCs. The black lines in for the size of
the catalyst.
As observed for Pt-Co catalysts in the previous section (Figure 4.8), no change in the
morphology or size of the NCs was observed following their addition to the carbon support and
their treatment to remove the ligands by O2 plasma and thermal annealing (232). The Pt-Ni NC
catalysts also exhibited good stability under HDO reaction conditions. Figure 4.20 shows TEM
images of the catalysts based on Pt6Ni NCs before (Figure 4.20a) and after (Figure 4.20b) 5 h of HDO
reaction. The thermal stability of NC catalysts was further confirmed by SAXS measurements
(Figure 4.20c). By subtracting out the scattering from carbon support (black curve) and performing
fitting with Rayleigh function, the average sizes of Pt6Ni/C before (green curve) and after (blue
curve) the functional testing were 4.1±0.8 and 4.1±0.7 nm, respectively. The results from SAXS
measurements indicate no change in the size of the NCs consistently with TEM observations. The
XRD patterns (Figure 4.20d) measured before (green curve) and after (blue curve) the catalyst was
exposed to the reaction environment also indicated that there was no phase transformation from
disordered fcc to ordered face-centered tetragonal (fct) structure. The carbon support was
measured as a control (black curve) and peaks near 25° and 43° correspond to graphite (002) and
diamond (111) planes for Vulcan carbon supports (233). The thermal stability of crystal structure
makes it possible to exclude a contribution from structural ordering on the catalytic performance.
150
Figure 4.20
The TEM images of 10 wt% Pt6Ni NCs on carbon support (a) before and (b)
after reaction. (c) SAXS patterns of Pt6Ni NCs in solution (red), 10 wt% Pt6Ni
NCs on carbon support before (green) and after (blue) reaction, carbon
support (black), and (d) the corresponding XRD patterns. The black lines in
(c) represent the simulated ts for the size of the catalyst.
Similar characterization studies were performed with the catalysts prepared from PtCu and
Pt2Zn NCs, with TEM images shown in Figure 4.21a-b. The addition of a stronger reducing agent
(BTB) resulted in smaller NCs than the ones prepared without BTB, which is consistent to what
previously reported (202). XRD patterns of the as-synthesized PtCu and Pt2Zn NCs also showed fcc
structure (Figure 4.21d). Based on SAXS data fitted to Rayleigh function (Figure 4.21c), the average
sizes and size distributions of the PtCu and Pt2Zn NCs are 6.6±1.2 and 2.6±0.6 nm, respectively. The
catalyst characterizations discussed above indicate that the NCs synthesized by solvothermal
methods were excellent subjects to study the relation between HDO selectivity and bimetallic
catalyst composition.
151
Figure 4.21
The TEM images of (a) PtCu and (b) Pt2Zn NCs, the corresponding (c) SAXS
patterns and (d) XRD patterns. The black lines in (c) represent the simulated
ts for the size of the catalyst.
4.4.3 HDO Experiments
In the previous section, the HDO reaction for HMF was shown to be a series reaction, with
products forming as shown in Scheme 4.2 (41, 185). Briefly, HMF is converted to a group of partially
hydrogenated intermediate species (B), which can react further to form DMF, which in turn may
react to over-hydrogenated products (D). Many monometallic catalysts, including carbon-
supported Pt, Pd, Ir, Ru, Co, and Ni, were found to exhibit relatively poor selectivity for DMF
formation (41, 185, 210). DMF yields varied with the particular metal catalyst and the reaction
conditions but were typically less than about 50%. The products formed by the further reaction of
DMF also varied with the metal catalyst, with Pt tending to form open-ring products, primarily 2-
hexanone and 2-propoxyhexanone (185).
Figure 4.22 shows a comparison of HMF conversions and DMF yields at 180 °C and 33 bar
total pressure for conventional 10 wt% Ni/C, Co/C, and Cu/C catalysts. Results for Ni/C and Co/C are
similar to what was previously reported (185). Both catalysts show high conversions, over 50%, for
even short residence times; and the DMF yield goes through a maximum of slightly more than 50%
with increasing space time, decreasing with longer space times. The detailed product distributions
152
are reported elsewhere (185), but the major over-hydrogenated product for both metals was 2,5-
hexandione. By contrast, the Cu/C was much less active, with a maximum conversion of less than
40% at the longest space time studied. The only products formed over the Cu/C catalyst were those
in the B group of Scheme 4.2, including MFu, HMMF and BEF. The low HDO activity observed over
Cu agrees with previous literature hydrogenation reactions of furfural over Cu catalysts, which
shows that the reaction over Cu tends to stop with formation of furfuryl alcohol [15]. Zn/C catalysts
did not show any activity for HDO of HMF under the conditions of this study.
Figure 4.22
Conversion and DMF yield for the HDO reaction of HMF over impregnated
10 wt% monometallic catalyst, as a function of reactor space time. Reaction
conditions: 33 bar and 180 ºC. () HMF conversion on Ni/C, () DMF yield
on Ni/C, () HMF conversion on Co/C, () DMF yield on Co/C, () HMF
conversion on Cu/C, () DMF yield on Cu/C. Zn/C catalysts did not show any
activity (not reported).
Similar experiments were performed over the Pt-Ni, Pt-Cu, and Pt-Zn bimetallic catalysts, in
order to determine how they would perform in comparison to previously discussed Pt-Co catalysts.
Figure 4.23 reports the HMF conversion and product yields as a function of space time for the three
Pt-Ni NCs/C catalyst at 160 °C and 33 bar. The results are qualitatively similar for all three Pt-Ni
catalysts, although the Pt3Ni NCs/C catalyst in Figure 4.23b) showed somewhat lower conversions.
At a space time of 0.5 g min mL-1, the conversion over Pt3Ni NCs/C was around 85%, while the
conversions on the other two Pt-Ni catalysts were over 95%. In general, HMF conversion increased
steadily with space time, while partially hydrogenated products (B group) yield diminished in favor
153
of DMF formation. Except for Pt3Ni NCs/C, DMF yield reached a value of 80-90% before decreasing
due to over-hydrogenated products (D group) formation. The fact that the D products form at the
same space times for which DMF yields begin to decrease suggests that they are formed from DMF.
The major D product observed with the Pt6Ni NCs/C catalyst was 2-hexanone, the same major
product formed on Pt catalysts (41, 185), while with PtNi NCs/C 2,5-hexandione was observed,
which was also the primary product formed on Ni/C catalysts (185). Interestingly, no over-
hydrogenated products were observed on the Pt3Ni NCs/C catalyst even for the highest space time
examined.
Figure 4.23
Conversion and product distribution for the HDO reaction of HMF over (a)
10 wt% Pt6Ni NCs/C, (b) 10 wt% Pt3Ni NCs/C, (c) 10 wt% PtNi NCs/C, as a
function of reactor space time. Reaction conditions: 33 bar and 160 ºC. ()
HMF conversion, () product group B, () DMF, () product group D.
Because the conversions for the Pt3Ni NCs/C were low at 160 °C, additional measurements
were performed on this catalyst at 200 °C, with results reported in Figure 4.24. At this temperature,
154
the conversions of HMF were much higher. Furthermore, DMF yields reached 98% at the higher
space times. It is not likely that the higher yields are due to the increase in temperature. In addition
to the fact that previous work indicated that selectivity are not very different between 100 and 200
°C (41, 210) one would expect the reaction of DMF to over-hydrogenated products to increase with
temperature. The maximum in the yield as a function of space time in Figure 4.24 is clearly less
steep than that found for Pt6Ni NCs/C and PtNi NCs/C catalysts. Furthermore, the Pt3Ni NCs/C
catalyst was remarkably stable. There was no observable change in either the conversion or the
DMF yield over a period of at least 5 h on the Pt3Ni NCs/C catalyst, while significant changes in both
were observed on Pt/C and Ni/C catalysts. The data in Figures 4.23 and 4.24 therefore indicate that
the Pt-Ni catalysts are more selective and stable than their monometallic counterparts and that
there is an optimum Pt:Ni ratio.
Figure 4.24
Conversion and product distribution for the HDO reaction of HMF over 10
wt% Pt3Ni NCs/C, as a function of reactor space time. Reaction conditions: 33
bar and 200 ºC. () HMF conversion, () product group B, () DMF, ()
product group D.
Catalysts based on Pt-Zn and Pt-Cu bimetallic NCs were also investigated, with results
shown in Figures 4.25 and 4.26. Because these catalysts were also less active than Pt/C, reactions
were carried out at 200 °C and 33 bar. The results for 10 wt% Pt2Zn NCs/C in Figure 4.25 are very
similar to results for Pt3Ni NCs/C, although the rates were slightly lower. Again, DMF yields as high
as 98% were achieved at high space time, with negligible production of over-hydrogenated
products. Similar performance was observed over the PtCu NCs/C catalyst, Figure 4.26, with DMF
yields again reaching 96%. However, alloying the Pt with Cu did significantly decrease the rates.
155
Figure 4.25
Conversion and product distribution for the HDO reaction of HMF over 10
wt% Pt2Zn NCs/C, as a function of reactor space time. Reaction conditions:
33 bar and 200 ºC. () HMF conversion, () product group B, () DMF, ()
product group D.
Figure 4.26
Conversion and product distribution for the HDO reaction of HMF over 10
wt% PtCu NCs/C, as a function of reactor space time. Reaction conditions: 33
bar and 200 ºC. () HMF conversion, () product group B, () DMF, ()
product group D.
156
4.4.4 Discussion
The results of the present study clearly demonstrate that the catalysts formed by alloying Pt
with a number of metals can significantly increase the yields for the HDO reaction of HMF to DMF.
In all cases, the increased yields result from a decreased reactivity of DMF to over-hydrogenated
products. The alloying metals appear to do this in different ways, but the net result in each case is
that the alloying metal prevents the furan ring from lying down on the catalyst surface. With Pt-Co
- x
overlayer on the Pt (210). Since Ni is more easily reduced than Co, the effect of Ni on the bonding of
the furans on the Pt-Ni alloys is likely due to the oxophilicity of the Ni, similar to what was reported
for the effect of Fe on Ni-Fe alloy catalysts (38). With Cu and Zn, the role of the alloying metals may
be to change the Pt ensemble size or otherwise change the way the DMF bonds to the metal
surfaces (224, 226).
The selective alloy catalysts were also more stable. In the present study, there was no
observable deactivation of a Pt3Ni/C catalyst, consistent with what was also observed for a Pt3Co2
catalyst (210). The results corroborate the statement that deactivation by coking in this reaction is
caused by the over-hydrogenated products such as di-ketones, which tend to be highly reactive.
With Pt-Ni and Pt-Co alloys, compositional uniformity of the catalyst is critical.
Monometallic catalysts based on Pt, Ni, and Co are not selective because all three metals will
catalyze the reaction of DMF to over-hydrogenated products (185). Furthermore, the selectivity of
the Pt-Ni and Pt-Co bimetallic catalysts depends on the Pt:Co (210) and Pt:Ni ratios. Synthesis of
bimetallic catalysts by conventional methods in which the support is infiltrated with metal salts is
not able to produce this uniformity, which is the reason the catalysts in this study were prepared by
synthesizing uniform NCs in solution. With Pt alloys of Cu and Zn, the uniformity will be less
important, given that Cu and Zn are not active for the reaction of DMF. In this sense, alloy catalysts
based on Cu and Zn could be easier to synthesize.
The fact that various Pt alloys show good selectivity raises a number of interesting
questions. First, would other alloys, including ones that do not include noble metals, also show
high selectivity? Second, can one generalize the high HDO selectivity for HMF that is observed with
the Pt alloys to HDO reactions with other reactants, such as those one might associate with lignin?
These will be interesting questions to address in the future.
157
4.4.5 Conclusions
Carbon-supported, monometallic Pt and Ni catalysts exhibit relatively low selectivity, less
than 50%, for DMF in the HDO reaction of HMF because DMF is converted to ring-opened or ring-
hydrogenated byproducts. On the other hand, Zn and Cu catalysts are not active for HDO reactions.
However, catalysts based on bimetallic Pt alloys with Ni, Zn, or Cu show significantly higher
selectivity to DMF, up to 98%. With Pt-Ni alloys, controlling the local composition is critically
important for achieving high DMF yields due to the non-selectivity of the pure metals. Even though
the effects of Ni, Zn, and Cu on Pt are expected to be very different, it is likely that each of the alloys
modifies the bonding of DMF so as to prevent the furan ring from lying down on the surface. This
appears to be the critical factor in achieving high selectivity for HDO of HMF.
158
4.5 Ni-Cu Alloys for HDO of HMF
4.5.1 Introduction
As discussed in the previous sections, bimetallic catalysts based on Pt have shown
significantly improved HDO selectivity to DMF with respect to single metal catalysts. It was
demonstrated for Pt-Co alloys that the reactivity of DMF was greatly suppressed on the alloy
compared to monometallic Pt or Co, due to formation of a CoOx monolayer that dramatically
weakened the interactions between the furan ring and the catalyst surface, so that over-
hydrogenation of DMF was suppressed (210). However, other Pt/base-metal (Pt-Cu, Pt-Zn) alloy
catalysts were also shown to give high yields of DMF due to suppressed reactions of DMF (>95%)
(232); and the alloying metal in the case of Cu is not as oxophilic as Fe or Co. It remains unclear
whether an oxide overlayer is needed in order to obtain high DMF yields.
Base-metal alloys have also been shown to be more selective relative to the monometallic
catalysts for the vapor-phase reaction of furfural to methyl furan. For example, Resasco and
coworkers have demonstrated that Ni-Fe catalysts are superior to monometallic Ni(38). In the
vapor-phase reaction, decarbonylation of furfural to furan is the major side product but there is
evidence that catalysts which are selective for HMF to DMF are also selective for furfural to methyl
furan (234). Higher yields of DMF from HMF have been reported for Ni-W2C (235) and Cu-Co@C
(236) catalysts, but there is no indication about the catalysts surface composition under reaction
conditions or about the role of the carbon overlayer in controlling the HDO selectivity.
In the present section is reported the reaction of HMF to DMF over Ni-Cu NCs catalysts in a
continuous flow reactor at high H2 pressures. DMF yields as high as 98.7% were achieved with
NiCu3 NCs catalyst due to the low catalytic reactivity towards DMF over-hydrogenation. Catalysts
prepared by conventional infiltration methods were not as selective as the ones prepared from
alloy NCs because the composition of the conventional supported metals was not as uniform. The
preparation of alloy NCs via solvothermal methods eliminates catalyst heterogeneity, which also
makes it easier to understand how catalyst composition and particle size affect activity and
selectivity (39). Near Ambient Pressure X-ray photoelectron spectroscopy (NAP-XPS) revealed that
the Ni-Cu NCs are completely reduced to the metallic state under reducing conditions. Possible
reasons for the high selectivity of this bimetallic alloy catalyst will be discussed.
159
4.5.2 Catalyst Synthesis and Characterization
NiCu and NiCu3 NCs were synthesized by a solvothermal synthesis under inert atmosphere.
To prepare the nanocrystals, nickel (II) acetylacetonate (Ni(acac)2 , 95%, Alfa Aesar) and copper (II)
acetylacetonate (Cu(acac)2 , 97%, Alfa Aesar) were dissolved in 20 mL of OAm (technical grade 70%,
Sigma Aldrich) and degassed at 100 °C for 15 min. Then, under N2 atmosphere, 1.5 mL of TOP (90%,
Sigma Aldrich) was added and the system was quickly heated to 230 °C. After 10 min, the reaction
mixture was cooled to room temperature. For the work-up, methanol was used to precipitate the
NCs by centrifugation (3 times), and hexane was used to redisperse the NCs after each step. In
order to obtain 1:1 and 1:3 Ni:Cu molar ratio, 3:1 and 1:1 molar ratios were used. The NCs were
dispersed into the carbon support to prepare 10 wt% of metals on carbon. To remove the organic
ligands from the surface of the NCs, we followed previously published procedures (210, 232).
Briefly, 100 mg of 10 wt% metal/C catalysts were exposed to O2 plasma for 15 min. After the plasma
treatment, the NC surfaces were further cleaned by rapid thermal annealing (RTA).
Conventional monometallic and bimetallic catalysts were synthesized by incipient wetness
impregnation for comparison with the nanocrystal catalysts. The Ni/C and Cu/C catalysts were
prepared with a 10 wt% metal loading. The metal precursors were nickel(II) nitrate hexahydrate
(98%, Alfa Aesar) and copper(II) nitrate trihydrate (98%, Alfa Aesar). Each metal precursor was
dissolved in a pre-mixed water/ethanol (3:1) solvent under vigorous stirring. Carbon black (Vulcan
XC-72R) was slowly introduced until uniform slurry was formed. The carbon support used in this
work has a BET surface area of 250 m2 g-1, a typical bulk density of 96 kg m-3 and an average particle
size of 50 nm. The wet mixture was then dried in air at 60 °C for 12 h, followed by 500 °C reduction
using 5% H2/He flow at 60 mL min-1. The temperature was programmed to reach 500 °C using a 3 °C
min-1 ramp rate and was maintained at 500 °C for 2 h before cooling. The conventional bimetallic
NiCu/C catalyst was prepared by co-impregnation, using a pre-mixed solution of the metal
precursors.
TEM images of the NiCu and NiCu3 NCs synthesized by solvothermal methods are shown in
Figure 4.27a and 4.27d. In each case, the particles were uniform, with a slightly faceted
morphology. The average sizes and size distributions of the NiCu and NiCu3 NCs were 13±1.0 nm
and 15±1.2, respectively. The slight difference in the particle sizes for NiCu and NiCu3 NCs is due to
the amount of precursors used during the synthesis. The XRD patterns in Figure 4.27g demonstrate
that the NCs have a face-centered-cubic (fcc) crystal structure. The diffraction peaks for NiCu NCs
were shifted to slightly higher angles compared to that of the NiCu3 NCs, as expected for well-
mixed alloys. An extra peak between (111) and (200) planes was observed for NiCu NCs, and this is
likely indicative of the presence of hcp-like stacking faults (237, 238).
160
Figure 4.27
The TEM images of (a) NiCu, (d) NiCu3 and the corresponding size
distribution (b) and (e) and diffraction patterns (c) and (f). The powder X-ray
diffraction patterns are shown in (g). Scale bars are 100 nm. Model colors:
light blue=Cu atoms, green=Ni atoms.
For SEM and NAP-XPS investigation, the NiCu and NiCu3 nanocrystals were deposited on a
graphite foil. SEM micrographs (Figure 4.28) showed that the nanoparticles were homogeneously
distributed over the surface in a side-by-side fashion, forming a single discontinuous layer. No
evidence for second (or higher) layer formation was observed. XPS spectra taken after treating the
NCs with oxygen plasma and Rapid-Thermal Annealing (RTA) showed that the pretreatment
effectively removed the protecting ligands, so that both N and P signals were below the detection
limit in the spectra (not shown).
161
Figure 4.28
SEM micrographs of graphite supported NiCux nanoparticles. (a) NiCu
nanoparticles after deposition, (b) NiCu nanoparticles after plasma
treatment and annealing to 700 °C. (c) NiCu3 nanoparticles after deposition,
(d) NiCu3 nanoparticles after plasma treatment and annealing to 700 °C. Size
of the images is 1 × 1 m2. Model colors: light blue=Cu atoms, green=Ni
atoms, red=oxygen atoms.
In order to understand the structure and chemical state of the NCs particles under reaction
conditions, we performed NAP-XPS characterization for both the NiCu and NiCu3 NCs, with spectra
shown in Figure 4.29. Immediately after evacuating the samples at room temperature, the spectra
of Cu and Ni showed that both elements were mostly oxidized. For Ni, the XPS signals can be
assigned to Ni2+ (853.7 eV), indicative of NiO, and Ni3+ (855.6 eV), indicative of Ni(OOH) or Ni2O3.
While the Cu signal has significant contributions from Cu2+ (934.7) and metallic Cu0 (932.4 eV), the
presence of Cu1+ cannot be ruled out due to its overlap with the metallic signal.
162
Figure 4.29
NAP-XPS spectra of Cu 2p and Ni 2p core levels of NiCu and NiCu3
nanoparticles measured in UHV and under 1 mbar of H2 at 250 °C. The
spectra are normalized to the metallic Cu 2p3/2 signal and offset for clarity.
XPS colors: light blue=Cu (0), green = Ni (0), red=oxidized metals,
grey=satellite signals; model colors: light blue=Cu atoms, green=Ni atoms,
red=oxygen atoms.
Spectra obtained in 1 mbar of H2 at 250 °C are also shown in Figure 4.29 and are more
indicative of the surfaces under reaction conditions. The reduction treatment led to complete
reduction of both Cu and Ni to metallic states in both the NiCu and NiCu3 NCs. Based on
photoionization cross-section analysis, both the NiCu and NiCu3 NCs exhibit a Ni:Cu molar ratio
close to 1:1 in the surface region mapped by XPS. This result implies that Ni segregates to the
surface in the NiCu3 nanoparticles, leading to the formation of a Ni-rich shell on a Cu-rich core. The
formation of such a core-shell structure is not surprising. During the sample preparation, the NCs
were mildly oxidized by oxygen-plasma and rapid-thermal-annealing treatments in order to
163
remove the surface organic ligands. Since Ni is easier to be oxidized than Cu, there is a driving force
to segregate Ni on the particle surface. After reduction, Ni forms an alloy with Cu, resulting in a Cu-
rich core and Ni-rich shell structure. Similar observations have been reported for Pt-Co. However,
with Pt-Co NCs spectroscopic data suggested formation of a stable CoOx layer at the surface that
could not be reduced under the conditions of that study (210).
4.5.3 Catalytic HDO Study
As previously discussed, high-pressure HDO of HMF in alcohol solvents is a sequential
reaction in which DMF is an intermediate product that can go on to form secondary products, as
shown in Scheme 4.2(41, 185). The product distributions as a function of reactor space time over 10
wt%, monometallic Ni/C and Cu/C catalysts are shown in Figure 4.30 for reaction at 180 °C and 33
bar total pressure. Results for Ni/C, shown in Figure 4.30(a), are similar to what has been reported
previously and are consistent with the sequential reaction scheme discussed above (41, 185).
Partially hydrogenated products (B) are formed only at the lowest space times and these rapidly
decline. DMF production initially increases with space time, then declines, while the over
hydrogenated products are formed only at longer space times. The maximum DMF yield was about
55%.
The Cu/C was much less reactive, as shown in Figure 4.30b. The maximum conversion of
HMF was less than 40% at the longest space time and the products were primarily the partial-
hydrogenated compounds (B). No DMF was observed on the Cu catalyst under the conditions
shown here. The results are consistent with the fact that Cu-based catalysts are used to
hydrogenate furfural to furfural alcohol (226, 227). Although the 1-propanol that we used as the
solvent in the present study can serve as hydrogen donor in catalyzed transfer hydrogenation
(CTH), we did not observe the formation of propanal in the products. Considering the fact that CTH
is commonly catalyzed by Lewis acid catalysts (239), the transfer hydrogenation likely did not
interfere the HDO reaction in this work.
164
Figure 4.30
Conversion and product yield for the HDO reaction of HMF over 10-wt%
impregnated (a) Ni/C, (b) Cu/C, as a function of reactor space time. Reaction
conditions: 33 bar and 180 ºC. () HMF conversion, () product group B,
() DMF, () product group D, () 5-methyl furfural (MF), () 2,5-
bis(propoxymethyl)furan (BEF), () 2-hydroxylmethyl-5-methyl furan
(HMMF).
Similar continuous-flow reaction experiments were carried out over the NiCu and NiCu3
bimetallic catalysts prepared by conventional impregnation method. Figure 4.31 presents data for
HMF conversion and product yields as a function of space at 180 °C and 33 bar for 10 wt% IMP-
NiCu/C and IMP-NiCu3/C catalysts. The product curves for IMP-NiCu, Figure 4.31a, and IMP-NiCu3,
Figure 4.31b, are qualitatively similar to the results shown earlier for the monometallic Ni catalyst,
but with different yields. Surprisingly, rates on all three Ni-containing catalysts were similar as
shown by the fact that conversions at any given space time were similar. For example, at a space
time of 0.5 g min mL-1, the conversion over IMP-NiCu/C was 85% and 60% on both NiCu3 and Ni
catalysts. This relatively small difference could be due to differences in metal dispersions, which are
difficult to measure on Ni-Cu alloys in general (240), especially when using a carbon support.
165
Figure 4.31
Conversion and product distribution for the HDO reaction of HMF over 10
wt% (a) IMP-NiCu/C, (b) IMP-NiCu3/C, as a function of reactor space time.
Reaction conditions: 33 bar and 180 ºC. () HMF conversion, () product
group B, () DMF, () product group D.
Both the NiCu and NiCu3 catalysts exhibited greatly improved DMF selectivity, with
maximum yields in the range of 80 to 90%. However, at longer space time, the DMF yields on both
Ni-Cu catalysts decreased, an indication that DMF was reacting to form over-hydrogenated
products (D group). The main over-hydrogenated products on the Ni-Cu catalysts were 2,5-
hexandione and DMTHF, which were also the primary products formed by reaction of DMF on Ni/C.
Unlike the reaction of HMF on Pt (185), the etherification by-products, 2-propoxyhexane or 2,5-
dipropoxyhexane, were formed in small amounts on the Ni-based catalysts.
In previous HDO studies of Pt-Ni and Pt-Co catalysts, compositional uniformity was shown
to be a critical factor in determining selectivity (210, 232); and NCs catalysts prepared by
solvothermal methods with the optimal composition showed greatly improved yields compared to
catalysts having the same overall composition but prepared by conventional impregnation. To
determine whether this would also be true with the Ni-Cu catalysts, we investigated the HDO
reaction on the highly uniform, 10 wt% NiCu and NiCu3, NCs catalysts, with conversions and yields
shown in Figure 4.32. Both NCs catalysts achieved better selectivity to DMF compared to
impregnated catalysts, with yields of 96.1% for the NiCu NCs/C catalyst and 98.3% for the NiCu3
NCs/C catalysts.
166
Figure 4.32
Conversion and product distribution for the HDO reaction of HMF over 10
wt% (a) NiCu NCs/C, (b) NiCu3 NCs/C, as a function of reactor space time.
Reaction conditions: 33 bar and 180 ºC. () HMF conversion, () product
group B, () DMF, () product group D.
Some deactivation of the HDO activities was observed over the Ni-Cu NCs. For example, at
180°C, 33 bar and 0.5 g min mL-1 W/F, the HMF conversion after 4 h decreased of about 10% over
NiCu NCs and nearly 50% over a conventional Ni catalyst. However, selectivity was not affected by
time-on stream. In previous work, it was argued that the deactivation is due to coking and blocking
of the metal active sites, based on the fact that stability scaled with the carbon balances on various
catalysts (210).
4.5.4 DMF Feeding Study
Because HDO of HMF is a sequential reaction in which DMF can undergo additional
reaction, high selectivity for DMF require that DMF be unreactive. Therefore, the reaction of DMF
over the various Ni-Cu alloy catalysts was investigated by feeding DMF in a solution with 1-
propanol at the same molar concentration as that used in the HMF reaction studies. The
experiments were carried out under the same reaction conditions, 180 °C with 33 bar H2; and the
data are summarized in Figure 4.33. The most notable difference between the catalysts was in their
activity, following the order: IMP-NiCu/C > IMP-NiCu3/C > NiCu NCs/C> NiCu3 NCs/C.
167
Figure 4.33
Conversion and product distribution for the reaction of DMF over 10 wt% (a)
IMP-NiCu/C, (b) IMP-NiCu3/C, (c) NiCu NCs/C, (d) NiCu3 NCs/C, as a function
of space time. Reaction conditions: 33 bar and 180 ºC. () DMF conversion,
() DMTHF, () 2-hexanone, () 2,5-hexandione.
The data in Figure 4.33 are consistent with the sequential reaction mechanism and the
reaction data for HMF. There is a direct relationship between the catalyst selectivity and the
reactivity of DMF. Furthermore, the products from the reaction of DMF were primarily DMTHF and
2,5-hexanedione, which were also the primary products that formed from HMF on these catalysts
at longer space times. Since less water is formed in the reaction of DMF, the yields of hexanedione
tend to be lower with DMF.
168
4.5.5 Discussion
The results of the present paper confirm that alloy catalysts are capable of higher selectivity
to DMF for HDO of HMF. DMF yields above 95% have been achieved over Pt-Co, Pt-Ni, Pt-Zn, Pt-Cu,
and, now, over Ni-Cu alloy catalysts (210, 232). In each of these cases, high selectivity have been
shown to be related to suppressed reaction of DMF to form over-hydrogenated products. Ni-Fe (38)
and Co-Cu (236) alloys have also been reported to show high selectivity for HDO reactions.
Although suppressed reactivity of DMF has not been confirmed in those cases, it seems likely that
the Ni-Fe and Co-Cu systems would show this similarity as well.
The important question is why all of these alloys exhibit reaction properties that are so
different from their monometallic counterparts. It seems likely that the alloying metal in each case
prevents the furan ring from lying down on the catalyst surface. Previous studies suggested that
the alloying metal oxophilicity was the key to achieve such conditions (38, 210). A similar
explanation cannot apply to the Ni-Cu NCs of the present study since NAP-XPS analysis indicates
the particles have a completely reduced surface at 250 °C even at low H2 pressure.
The role of Cu in the Ni-Cu alloys may be to decrease the ensemble size at the Ni surface
(241). An alternative explanation is based on results from Ke et al. (242) who investigated the HDO
reactions using adsorption of furfural on Cu(111), Ni(111), and Ni-Cu model surfaces. On Cu(111),
1(O) configuration, whereas furfural was
2(C, O) configuration and decomposed non-
selectively. Selective reaction of furfural to 2-methyl furan was observed on a Cu(111) surface with
a Ni overlayer.
The differences between the Ni-Cu catalysts prepared by conventional impregnation and
those prepared from nanocrystals are likely due to catalyst homogeneity. The individual metal
particles in the impregnated catalysts almost certainly have varying compositions and the
presence of some Ni-rich particles will limit the selectivity. This inhomogeneity problem is
particularly serious when dealing with sequential reactions in which the desired product is an
intermediate. Even if the majority of the catalyst is selective, that part of the catalyst that is not can
further promote reaction of the desired product. Compared to alloys like Pt-Ni and Pt-Co, Cu-based
alloys will be less sensitive to compositional heterogeneity because Cu is relatively unreactive.
However, the advantages of the nanocrystal catalysts prepared by solvothermal methods were still
apparent with the Ni-Cu system.
The oxidation state of the catalyst surface during reaction is important to consider. Our
present XPS results demonstrate that the Ni-Cu catalysts are reduced with very mild pretreatments,
169
so that the catalyst is certainly reduced before reaction. A calculation of the bulk thermodynamics
for the reaction Ni +H2O = NiO + H2 show that equilibrium will occur at a P(H2) : P(H2O) ratio ~10-3 at
180 °C. Since the P(H2) : P(H2O) ratio was never less than 25 in the HDO reactor, the
thermodynamics of the nanoparticles would have to be dramatically different from that of the bulk
for the particles to be oxidized under reaction conditions. In previous work on Pt-Co NCs catalysts,
where a surface CoOx was observed(210), bulk thermodynamics would again suggest that the Co
should be in its metallic form. In addition to the fact that the thermodynamics of surface Co
reduction may be somewhat different from that of the bulk(243), it appears that reduction may
have been kinetically limited in that case, since the catalyst was not reduced, even by the initial
pretreatments.
The fact that Ni-Cu alloys are capable of providing high selectivity for DMF production
demonstrates that reasonably inexpensive catalysts can be used for selective HDO of HMF. It is
interesting to ask whether these base-metal alloys could also be used for other HDO reactions,
such as would be important in the processing of lignin. There are indeed indications for this (244).
4.5.6 Conclusions
Monometallic Ni and Cu catalysts exhibit low selectivity for DMF formation in the HDO
reaction of HMF due to over-hydrogenation on Ni and low reactivity on Cu. On the other hand,
bimetallic Ni-Cu catalysts show very high DMF yield in the same reaction conditions. The surface
composition and chemical state of the active phase were clearly identified by state-of-the-art NAP-
XPS analysis on Ni-Cu catalysts based on bimetallic nanocrystals of controlled dimension and
composition. In both nanocrystal NiCu and NiCu3-based catalysts, Ni and Cu were completely
reduced already at 250 °C in 1mbar of H2 and exposed surface in a 1:1 molar ratio, strongly
indicating that an oxide overlayer is not necessary to achieve high DMF yield in the HDO reaction
of HMF.
170
5. Dye-Sensitized Photocatalytic H2
Production
5.1 Introduction
H2 is an ideal energy carrier because it has no carbon footprint, it can be obtained from
water and it can be used as fuel in fuel-cells or as a reagent to produce fuels, biofuels (see Chapter
4) or chemicals. Nowadays, H2 is mostly produced by steam reforming of methane or other fossil
fuels, but many other technologies are available, such as biomass gasification and electrolytic
processes, or in development, such as photoelectrochemical (PEC) or biological processes (245). In
PEC devices H2 is produced from water (and usually some sacrificial agent), using sunlight and
specialized semiconductors called photoelectrochemical materials. PEC water splitting is a
promising pathway for H2 production at semi-central and central scales, but solar-to-hydrogen
efficiencies are still very low, and continued improvements in efficiency, durability, and cost are still
needed for market viability (246).
One reason for low efficiency is that water splitting is a multi-electronic, multi-atomic,
thermodynamically demanding and kinetically hampered process with a high activation barrier.
The standard potential ( E°) of water splitting or water electrolysis to H2 and O2 (Equation 5.1) is
1.23 V at any pH. Moreover, under operation the voltage required for water splitting is higher than
predicted (up to 1.8 2.0 V), due to thermodynamic losses and overpotentials associated with the
reaction kinetics (247). Another drawback is that hydrogen and oxygen are produced in the same
environment and easily recombine before they can be separated. Therefore, a common approach
for testing new materials is to limit the process to either one of the half-reactions: water reduction
to hydrogen or oxidation to oxygen, by the use of a sacrificial electron donor (SED) or acceptor
(SEA), respectively. SEDs and SEAs act as holes and electron scavengers, respectively, replacing
water in one of the two half-reaction.
(5.1)
Another reason for low efficiency of PEC devices is related to the semiconductor used and
its electronic properties. Titanium dioxide,TiO2 , is the most investigated and employed material in
171
both fundamental research and practical applications, because it is a cheap, nontoxic, chemically
and biologically inert, and photo-stable material; especially when coupled with a co-catalyst, such
as Pt, TiO2 performs very well in many photocatalytic reactions (42). However, TiO2 does not absorb
visible light, because of its large band gap (3.2 eV, depending on polymorph and nanostructure),
so that only a narrow, UV (Ultraviolet) portion of the solar spectrum can contribute to electron
excitation from the valence band (VB) to the conductive band (CB). Many methods have been
adopted in order to extend the light absorption of TiO2 in the visible (Vis) range, such as doping
with metals and non-metals (248), addition of plasmonic nanoparticles (Au in particular) (249),
reduction to black TiOx (250) and nanostructural engineering by preparation of nanocrystals (200).
Extended light absorption was shown to improve the catalytic performance of TiO2 , but it isn't
always effective for enhancing water splitting performance in PEC applications.
Coupling TiO2 with visible light-absorbing moieties such as colored dyes is an alternative
approach to enhance the photocatalytic efficiency of TiO2, largely employed in the field of solar
cells (i.e. dye-sensitized solar cells (DSSCs)) (32). Instead of changing the optical properties of the
semiconductor, a dye acts as a photosensitizer, or antenna, whose main purposes are to efficiently
absorb Vis light and to trigger the remaining steps of the water splitting process by electron
injection into TiO2 CB. Figure 5.1 illustrates the working mechanism of a dye-sensitized Pt/TiO2
photocatalyst for H2 production, summarized in Equations 5.2-5.5.
Figure 5.1 Graphical representation of the working mechanism for H2 production over
dye-sensitized Pt/TiO2 photocatalysts.
172
(5.2)
(5.3)
(5.4)
(5.5)
(5.6)
(5.7)
The dye absorbs a photon and enters an excited state (5.2). Electron injection into the CB of
TiO2 results in charge separation (5.3). Fast regeneration of the dye by oxidation of the SED agent
must take place for an efficient process and high durability of the dye (5.4). Electrons are then
transferred to Pt(0) nanoparticles adsorbed on the TiO2 surface, on which water (H+) is reduced to
molecular hydrogen (5.5). The ideal process can be hindered by some unwanted events. The two
main causes of loss in activity are reported in Equations 5.6-5.7: the relaxation of the dye to its
ground state before electron injection into the CB of the semiconductor (5.6) and hole electron
recombination between TiO2 and dye (5.7), which is in competition with H2 generation (5.5) and
dye regeneration (5.4) steps.
Most photocatalytic studies are performed in a suspension of dye-sensitized TiO2 particles
covered by Pt, which is deposited through impregnation or photodeposition from H2PtCl6 under
UV irradiation conditions (251). Dye staining is performed by suspending Pt/TiO2 powders in a dye
solution (usually, less than 1 g of powder in some mL of a mM solution of the dye) for a few hours in
the dark. The powders are then separated through centrifugation, washed and dried. In order to
estimate the amount of adsorbed dye, its concentration in the residual staining solution can be
ascertained by UV/Vis spectroscopy.
In order to test the dye/Pt/TiO2 catalytic activity, the material is suspended in an aqueous
medium containing the SED agent. Typical SEDs are triethylamine (TEA), triethanolamine (TEOA),
and ethylenediaminetetraacetic acid (EDTA), but also other inorganic (e.g., S2 , SO32 , Fe2+, Ce3+, I ,
Br , and CN ) and organic (alcohols, aldehydes, acids) SEDs have also been investigated (252). The
pH of the reaction mixture is typically adjusted to neutral. The photoreactor is then evacuated to
remove oxygen and irradiated with a Xenon lamp or a solar simulator provided with a UV filter at
ca. 400 410 nm in order to cut off the UV portion of the irradiation to avoid direct TiO2 excitation of
electrons to the CB of the semiconductor. The produced hydrogen gas is finally quantitatively
173
determined with a gas chromatograph equipped with a thermal conductivity detector (TCD). The
amount of hydrogen evolved over time and the relative rates of production (typically in the order
of µmol h-1) are usually normalized to the weight of the catalytic powder (µmol h 1 g 1) for practical
reasons or to the surface area of the semiconductor for a correct comparison between different
materials (253). Since the light absorption is correlated to the presence of dyes, the obtained values
of hydrogen production can be compared only among catalysts having similar loadings of dye and
co-catalyst. In order to evaluate and compare the performance of photocatalytic systems having
the same surface area, some specific parameters are used, such as TON, TOF and LFE which are
defined in Section 2.11.
Sensitizers for photocatalytic H2 production can be classified in three main categories:
organometallic complexes, natural or bio-inspired dyes, and metal-free organic dyes.
Organometallic complexes have been widely investigated in DSSCs applications, but in spite of the
high performances in DSSCs, their activity in H2 production is (with some exception) not
exceptional (254). The first class of organic dyes that was deeply investigated is that of emissive
dyes, but the donor acceptor molecular architecture represents a more general and interesting
approach for the design of organic sensitizers. A donor acceptor structure is composed of three
sub-molecular units: an electron-donor group (D), a -spacer ( ), and an electron-acceptor group
(A). The D A framework has often been used in material science because it is associated with
efficient charge separation. Donor acceptor dyes follow an oxidative quenching mechanism, in
which the donor is excited, injects an electron to the CB of TiO2 and the oxidized dye (dye+·) is then
regenerated by the sacrificial agent (46). After electron donation from the excited state, the hole of
the oxidized dye resides in the HOMO, localized in the D moiety. Therefore, the D group should lie
as far from the TiO2 surface as possible, in order to prevent charge recombination, and as close to
the SED as possible in order to favor dye regeneration. -Spacers are polarizable conjugated
aromatic and heteroaromatic groups able to efficiently transfer charge between D and A moieties
and to extend the -conjugated framework for improved optical properties and light harvesting.
The nature of the spacer unit plays an important role in stabilizing the dye under irradiation. The A
group is a strong electron-withdrawing moiety, able to promote charge separation in the excited
state. The molecular LUMO, from where electron injection into TiO2 takes place, is typically located
in the terminal A component. Therefore, the A group is covalently anchored to the semiconductor
surface by a functional group, to keep it in close proximity to the catalytic center (46).
Careful molecular design of the sensitizer is of fundamental importance to develop efficient
and stable hydrogen production photocatalysts. The efficiency of a dye depends in first place on its
optical properties. In order to be efficient, a sensitizer requires precise tuning of HOMO/LUMO
energetic levels (46, 251). In particular, the energy of the LUMO should be higher than the CB of the
174
semiconductor to allow electron donation, while dye regeneration can only take place if the HOMO
energy level of the dye is lower than that of the SED. HOMO and LUMO energies can be
experimentally determined electrochemically (e.g. Cyclic Voltammetry ─ CV) or optically (Tauc
plots) and their shapes and energies can be calculated via ab initio computations, mostly based on
the DFT method (46). Many other factors influence the performance of a dye-sensitized
photocatalyst, such as dye stability upon irradiation, wettability of the semiconductor surface by
water after staining (dictated by ancillary functionalities of the dye), anchoring unit stability and
injection efficiency (46).
The most widely investigated donor acceptor sensitizers for dye-sensitized photocatalytic
H2 production are based on triarylamines (TAA) and phenothiazines (PTZ) D moieties (46) (Figure
5.2). The PTZ core has particular features associated with its non-planar butterfly conformation
along the S N axis. This arrangement helps to minimize the negative effects associated with self-
quenching molecular aggregates on the TiO2 surface. PTZ contains two symmetric benzene rings
which can be conveniently functionalized, allowing the design of symmetric di-branched dyes.
Such dyes present high anchoring stability and electron injection efficiency, improved optical
properties, and enhanced device stability (255). Furthermore, the nitrogen atom of the central
heteroaromatic ring can be conveniently functionalized in order to tune additional properties, such
as water solubility. Finally, in comparison with other donor groups, the PTZ ring radical cation form
is very stable, thereby facilitating electron donation and regeneration by SED.
Figure 5.2
Typical PTZ-based sensitizers structure: the N-functionality (R) can be an
alkyl chain of variable length, or a hydrophilic moiety. The spacer units,
here for di-branched dyes, are usually composed of one or more
functionalized or simple heterocyclic groups (e.g. thiophene, 2,2'-
bithiophene, 1,4-thiophthene). The most common anchoring group
(cyanoacrylic) is also depicted.
175
In the following sections, we report the rational molecular design of donor-acceptor dyes
based on the PTZ D core aimed to enhance the H2 production performances of Pt/TiO2 sensitized
photocatalysts. In particular, the effect of peripheral functionalization (aimed to increase surface
wettability) and heteroatom substitution in the D core and π-spacers will be discussed. Finally,
some future perspective on H2 sustainable photocatalytic production from EtOH/water mixtures
will be presented.
176
5.2 Phenothiazine-based Sensitizers: N-functionalization Effect
5.2.1 Introduction
In this section, we investigate the effect of hydrophilic substituents of the PTZ moiety of a
dye-sensitized TiO2 photocatalyst (TiO2-NP) on the photocatalytic hydrogen production from
aqueous solutions. One of the most common strategies to induce hydrophilicity to organic dyes is
the introduction of a polyethylene glycol functionality, such as the widely used tris(ethylene glycol)
monomethyl ether (TEG) group (256). More recently, the use of poly-glycolic functionalities as
substituents in organic molecules for dye-sensitized hydrogen generation has been also reported
(257). In this study, TEG (PTZ-TEG), sugar methyl -D-glucopyranoside (PTZ-GLU) and (n-octyl)
(PTZ-ALK) derivatives of a thiophene-based PTZ dye were synthesized, characterized and tested
for sensitization of Pt/TiO2 (Figure 5.3).
Figure 5.3 PTZ-based sensitizers investigated in this work.
The multifunctional scaffold methyl -D-glucopyranoside was employed in order to further
tune water affinity of sensitizers and to induce additional properties such as intermolecular self-
assembly. Carbohydrates are excellent hydrogen bond donors and acceptors because of their
multiple hydroxyl functions and can therefore induce self-assembly (258). The unprecedented use
of a sugar functionalization in this field was demonstrated to improve the wettability of TiO2
177
materials after staining, compared to both PTZ-ALK and PTZ-TEG. The sunlight-driven H2
production can be related to the chemical structure of the dyes and to the distinct features of the
side functionalities, both in terms of evolution rates and TON.
5.2.2 Catalyst Synthesis and Characterization
In order to modify the wettability properties of the catalyst surface without interfering with
the conjugated skeleton of the sensitizer (which is responsible for the light harvesting step of the
photocatalytic process), the hydrophilic groups were introduced on the terminal electron-rich D
core of PTZ by exploiting the nitrogen site of the heterocyclic ring. A methylene linker was used in
order to block communication with the system. In this way no modification of on the -
conjugated system was induced and the effect of the terminal groups on H2 production was only
related to wettability of the sensitized photocatalyst. The sugar functionality of PTZ-GLU has been
introduced by exploiting click chemistry and Cu-assisted azide alkyne Huisgen cycloaddition (259,
260). Although alternative synthetic paths could be envisaged, click chemistry approach was
chosen in order to ensure design flexibility, extension to a library of glycoconjugated dyes
(261), and complete transfer to the industrial scale. The sugar is in form of -methyl
glucopyranoside in order to prevent any possible redox interference of the anomeric free aldehyde
with the photocatalytic cycle of the sensitizer. For details about dyes synthesis procedure and
characterization, see Ref. (262).
The absorption spectra of dyes PTZ-ALK, PTZ-TEG, and PTZ-GLU are reported in Figure 5.4
and the main optical and energetic (HOMO LUMO energies) parameters are listed in Table 5.1. As
expected, the optical properties of the three dyes were not significantly affected by the nature of
the functionality. The three dyes showed a typical intense * absorption band in the Vis region
attributed to the intramolecular charge-transfer transition. The position of the absorption
maximum is about 470 nm whereas peak molar absorptivity is slightly different for the three dyes.
178
Figure 5.4 Absorption spectra of dyes PTZ-ALK, PTZ-TEG and PTZ-GLU in THF
Table 5.1 Optical and electrochemical characterization of dye PTZ-ALK, PTZ-TEG and
PTZ-GLU. PTZ-ALK values from Ref. (262); for CV and DPV plots: Ref. (262).
Contact angle analysis was used to investigate the hydrophilicity properties of
the sensitized TiO2 nanoparticles. The contact angles of a DI water drop on the surface of a film of
sintered TiO2 and the corresponding dyes-sensitized films are shown in Figure 5.5. Data are
summarized in Table 5.2. The bare TiO2 and the hydrophilic dyes-sensitized films have contact
angles lower than 35°, compared to 117° for the hydrophobic dye PTZ-ALK. The sugar functionality
in PTZ-GLU is able to improve water affinity by further decreasing the angle from 34° in the PTZ-
TEG dye to 27°, thus going closer to the bare TiO2-NP surface character.
179
Figure 5.5 Cross-sectional images of a film of sintered TiO2-NP (a) and corresponding
samples sensitized with PTZ-ALK (b), PTZ-TEG (c) and PTZ-GLU (d).
Table 5.2
Contact angle and photocatalytic performance of the dye/Pt/TiO2 materials
in H2 production from TEOA 10% v/v solution at pH = 7.0 under irradiation
with visible light ( > 420 nm).
A Pt/TiO2 nano-powder was used as benchmark material to test the comparative
sensitization ability of the new synthesized dyes under irradiation with visible light ( > 420 nm).
The deposition of Pt nanoparticles on TiO2 Degussa P25 was done through a previously reported
photodeposition method (257, 263). 32.7 mg of Pt(NO3)2 was dissolved in 300 mL of a solution of
water/methanol 1:1 v/v. 2.0 g of TiO2 Degussa P25 were suspended in the Pt solution in order to
180
achieve a metal loading of 1.0 wt%. After stirring for 1 h in the dark, the suspension was irradiated
with a 450 W medium pressure lamp for 4 h. The Pt/TiO2 nanocomposites were collected by
centrifugation, washed 3 times with water and finally dried at 80 °C overnight. The TiO2 P25
support was an anatase/rutile mixture (∼70/30 by weight) with mean crystallite sizes of 20 nm for
both phases. Textural analysis revealed a surface area of 55 m2 g 1 with pores diameters around 48
nm and a pore volume of 0.242 mL g 1. HAADF-STEM analysis (Figure 5.6) evidenced the irregular
shape of TiO2 particles (12 45 nm), with Pt nanoparticles with mean size of 2.4 nm homogenously
distributed on the surface of the support.
Figure 5.6 Representative HAADF-STEM images of the Pt/TiO2 nanocomposite.
The PTZ-based dyes (PTZ-ALK, PTZ-TEG and PTZ-GLU) have been adsorbed on
Pt/TiO2 nanocomposite by suspending 200 mg of Pt/TiO2 nanocomposite in 10 mL of dye solution
in ethanol for 24 h in the dark. The concentration of the dye was adjusted in order to obtain the
desired loading, between 1.0 and 30.0 μmol g-1. The obtained materials were recovered by
centrifugation, washed 2 times with 10 mL EtOH each and dried under vacuum at room
temperature overnight. UV-vis spectra of the solutions after dye adsorption showed that the
amount of residual dyes is negligible, confirming the quantitative adsorption of the dyes on
Pt/TiO2.
181
5.2.3 H2 Photocatalytic Production Experiments
The Pt/TiO2 photocatalysts sensitized by PTZ-ALK, PTZ-TEG and PTZ-GLU were tested for
H2 production under Vis light ( > 420 nm) from a triethanolamine (TEOA)/HCl aqueous buffer
solution at pH = 7.0. Following the "best practice in photocatalysis" reported by Kisch and
Bahnenmann (45), the experimental conditions have been optimized measuring the H2 production
rate after stabilization (see below) using different amounts of the photocatalyst. This preliminary
optimization was performed using the PTZ-GLU/Pt/TiO2 photocatalyst with a dye loading of 30
mol g 1 (Figure 5.7). Maintaining constant all the other experimental factors (geometry of
irradiation and reactor, volume of the TEOA/HCl solution, temperature, etc.), the maximum
H2 production rate has been obtained using 60 mg of the photocatalyst, with a slight decrease for
higher amounts likely due to increased scattering of the incoming photons.
Figure 5.7 Optimization of the experimental conditions for photocatalytic H2
production.
For all the sensitized photocatalysts investigated in this study, the H2 production rates
increased in the first hours of irradiation until reaching a constant mean value (Figure 5.8). This
phenomenon arises from the combination of two factors. First, diffusion of produced H2 in the
dead volume of the photoreactor resulted in the progressive increase of H2 concentration in the
gaseous effluent for the system, which typically accounts for the first 1 2 h of time-on-stream.
Second, activation of the photocatalyst took place at the beginning of each photocatalytic
experiment, likely because the Pt nanoparticles were passivated by the adsorbed oxygen resulting
182
from exposure to air after photodeposition. The activation period is significantly lower for the
hydrophilic dye, likely because of the better interaction with the aqueous reactants. Notably, no
H2 production was observed using the bare Pt/TiO2 under the same experimental conditions.
Moreover, the contribution of degradation of the glucose group in the photocatalytic process can
be reasonably ruled out: a PTZ-GLU/Pt/TiO2 catalyst (loading = 10 mol g 1) without the sacrificial
donor TEOA afforded a H2 production rate below the instrument detection limit (∼3 mol
H2 g 1 h 1).
Figure 5.8
H2 production rates from TEOA 10% v/v solution at pH 7.0 under irradiation
with visible light ( > 420 nm) measured over the dye/Pt/TiO2 materials.
Numbers in the legends correspond to the dye loading in mol g-1.
183
The comparison of the H2 production rates as a function of the dye loading showed
different behavior for the photocatalysts sensitized with the investigated dyes (Figure 5.9). The
effect of dye loading on the performance of photocatalysts with PTZ-ALK and PTZ-TEG shows the
same trend, although the hydrophobic dye afforded a better performance. A sharp increase in the
activity is observed for dye loadings below 2.5 mol g 1, while for higher loading the activity
increment is modest. The photocatalysts sensitized by PTZ-GLU showed a different trend, with the
H2 productivity increasing up to the loading of 10 mol g 1.
Figure 5.9
Production rates in H2 evolution from TEOA 10% v/v solution at pH = 7.0
under irradiation with visible light ( > 420 nm) using the Pt/TiO2 materials
sensitized with PTZ-ALK, PTZ-TEG and PTZ-GLU.
The different behavior of the dyes in sensitizing Pt/TiO2 photocatalysts can be rationalized
in terms of their structure. The comparative trends are in agreement with the similar molecular
geometry of PTZ-ALK and PTZ-TEG, in which a linear and flexible terminal chain is present. In the
case of PTZ-ALK, the chain is reasonably coiled up because of repulsion from the aqueous solution,
suppressing intermolecular quenching. This phenomenon does not take place in the case of PTZ-
TEG, where the polar group is likely unrolled allowing interaction between the heteroaromatic
units.
The PTZ-GLU sensitized photocatalysts showed H2 production rates comparable to PTZ-
ALK for high loadings (>10 mol g 1), while at small loadings the rates were much lower. PTZ-
GLU has a bulky side chain, with a considerably smaller degree of freedom compared to the ALK
and TEG chains. In fact, whereas a rigid cyclic substituent is present for PTZ-GLU, in PTZ-
184
ALK and PTZ-TEG the side chains are endowed with free rotation along single bonds. This
suggests a distinct arrangement of PTZ-GLU on the surface of Pt/TiO2. At high loadings, the
organization of PTZ-GLU becomes somewhat more similar to that of PTZ-ALK, with the PTZ units
interacting with the Pt/TiO2 surface and the bulky lateral chains avoiding intermolecular
quenching. Notably, the glucose functionality could also induce some sort of supramolecular
organization on the surface (258, 264). For low loading of PTZ-GLU, the glucose unit could interact
directly with the TiO2 surface through the remaining OH groups and this might change the
orientation of the PTZ scaffold, affecting the electron transfer to TiO2.
TON values and Light-to-Fuel Efficiency (LFE20) measured after 20 h of irradiation (Figure
5.10 and Table 5.3) showed that the efficiency of light conversion into effective electrons to
H2 increased as the dye loading decreased, in agreement with the aforementioned discussion
relating the molecular structure to the photocatalytic activity.
Figure 5.10
TON in H2 evolution from TEOA 10% v/v solution at pH = 7.0 under
irradiation with visible light ( > 420 nm) using the Pt/TiO2 materials
sensitized with PTZ-ALK, PTZ-TEG and PTZ-GLU.
185
Table 5.3 TON values and Light-to-Fuel Efficiencies (LFE) of dye PTZ-ALK, PTZ-TEG
and PTZ-GLU.
5.2.4 Conclusions
In conclusion, a sugar derivative of a multi-branched organic sensitizer (PTZ-GLU) has been
efficiently used in the dye-sensitized photocatalytic production of H2 through careful catalyst
design. The insertion of the glucose unit yielded a higher affinity towards the aqueous medium
compared to the commonly used hydrophilic TEG functionality, while maintaining the high activity
recorded for the alkyl derivative. On the basis of contact angle measurements, photocatalytic data,
and the peculiar structural features of the side substituents, the distinct behavior of PTZ-GLU was
ascribed to the rigid, bulky, hydrophilic geometry of the glucose ring. The lower degrees of
freedom and the extra-wettability of the PTZ-GLU dye favored the interaction with the reactants in
aqueous solution, suppressing intermolecular quenching. The general and scalable synthetic
approach can be adapted to a large variety of sugar derivatives, to synthesize photosensitizers
having finely tuned properties, and to study their effect on the efficiency of H2 photocatalytic
production.
186
5.3 Carbazole-based Sensitizers
5.3.1 Introduction
In this section, we report a rational investigation of the effect of heteroatoms replacement
(or removal) in a previously reported PTZ dye on the H2 production over sensitized Pt/TiO2
photocatalysts. In particular, new sensitizers were prepared by replacing the PTZ donor core with
either a carbazole (CBZ) or a phenoxazine (POZ) moiety, and by replacing the thiophene spacers
with furan spacers. The CBZ-based dyes were demonstrated to be much more efficient as Pt/TiO2
sensitizers in the H2 photocatalytic production, compared to both PTZ and POZ based dyes. The
amount of produced H2 and turnover numbers (TON) are top-ranked amongst studies on dye-
sensitized photocatalytic hydrogen production (46).
CBZ-based systems are widely studied in the fields of materials science (44, 265, 266), and in
dye-sensitized solar cell (DSSCs) (267 269), only very few studies on photocatalysis have been
reported (270 272). Whereas PTZ has a bent, butterfly structure (273), CBZ is planar and contains a
5-member electron-rich heteroaromatic ring, with a pyrrole-like N atom (274). The planar structure
could enhance the photocatalytic efficiency of the sensitized Pt/TiO2 system because of favoured
charge generation and transport in the dye. In order to separate the effects of S atom replacement
and of dye structure, POZ donor groups (having a butterfly structure and an O atom instead of S)
were investigated. To study the role of sulfur in the spacer unit, new dyes were designed by
replacing thiophene (Th -spacers with furan rings (Fu), which have many electronic and
structural properties in common, with the exception of the nature of the ring heteroatom (274).
The structures of the dyes reported in this section are summarized in Figure 5.11.
Figure 5.11 The sensitizers investigated in this work and their structure.
187
5.3.2 Catalyst Synthesis and Characterization
The dyes investigated in this work were synthesized in the research group of Professor
Alessandro Abbotto (Bicocca University - Milan) by modifying the previously reported procedure
for the synthesis of PTZ-Th (PTZ-ALK in the previous section) (262). The same synthetic scheme
(Figure 5.12) was adopted for all the dyes. A detailed discussion on the synthesis is out of the
purpose of the present thesis. Briefly, Suzuki cross-coupling (Figure 5.12i) was adopted to react the
D-core organobromide derivative 1 with the desired π-spacer precursor (commercially available 2-
furanaldehyde-5-boronic acid for Fu dyes and 2-thiophenealdehyde-5-boronic acid for Th dyes). In
order to functionalize the dye with the anchoring unit, a Knoevenagel condensation was employed
(Figure 5.12 ii), starting from the dye aldehyde precursor 2, which was reacted with cyanoacetic
acid and piperidine to yield the final dye upon purification. Although the synthesis of CBZ-Th and a
molecule similar to CBZ-Fu (bearing a different N-alkyl functionalization) have been previously
reported in the literature (269, 275), the CBZ-based sensitizers were synthesized by the
aforementioned general approach using 3,6-dibromocarbazole as a starting reagent while POZ
derivatives were synthesized from 10-octyl-10H-phenooxazine-3,7-dibromo.
Figure 5.12
General synthetic procedure for dyes PTZs, POZs, CBZs. Reagents and
conditions: (i) 5-formyl-2-arylboronic acid, Pd(dppf)Cl2·CH2Cl2 [dppf =
1,1′-bis(diphenylphosphino)ferrocene], K2CO3, DME/MeOH, microwave
100 °C, 90 min; (ii) cyanoacetic acid, piperidine, CHCl3, reflux, 5 h.
188
The absorption spectra of the dyes (1.0 10-5 M in DMSO) are shown in Figure 5.13 and the
detailed main optical and energetic parameters (HOMO-LUMO energies) are listed in Table 5.4. In
general, the three families of PTZ, POZ and CBZ dyes exhibited 2 typical -
absorption in the 300 450 nm range and to the intramolecular charge transfer (ICT) transition in
the 400-600nm range (43, 269). - -shift in the order POZ-PTZ-
CBZ, as a result of increased electron delocalization in the core. On the other hand, the ICT
transition is subjected to a progressive blue- -
overlap for the CBZ derivatives. The absorption maxima are centered around ca. 470 nm for PTZ,
ca. 410 nm for CBZ and ca. 530 nm for the POZ dyes. The maximum molar absorptivity is higher for
the CBZ derivatives, whereas marginal differences (less than 10%) are recorded between PTZ and
POZ dyes. Finally, the introduction of the Fu spacer in place of Th did not significantly affect the
absorption properties.
Figure 5.13 Absorption spectra of the PTZs, CBZs and POZs dyes recorded in
DMSO solution.
189
Table 5.4 Main optical and electrochemical characterization of the PTZs, CBZs and
POZs dyes. Values for PTZ-Th from (262); for CV DPV plots see: Ref. (262).
Differential Pulsed Voltammetry (DPV) was used to determine the HOMO energy levels
from the current onset, while LUMO levels have been derived from electrochemical HOMO values
and optical band gaps, measured by means of Tauc plots (276). Levels are pictorially shown in
Figure 5.14. Even though the HOMO energy levels are quite similar for most of the dyes (~ -5.60 eV),
their LUMO energies and, accordingly, the electron injection capabilities to the Pt/TiO2 system are
different. In particular, the LUMO energy of the POZ dyes, as well as that of PTZ-Fu, are very close
to the conduction band (CB) of TiO2 (-4.0eV) (277).
190
Figure 5.14
Experimental HOMO/LUMO energy levels for the dyes investigated in
this work (CB level of TiO2 is included as a reference for electron
injection from the dye to the semiconductor).
Sensitized Pt/TiO2 nano-powder were prepared according to the procedure described in
Section 5.4.2. Briefly, Pt nanoparticles were photodeposited on TiO2 Degussa P25 and the dyes
were adsorbed on Pt/TiO2 nanocomposite by suspending the powder in a diluted dye solution. The
concentration of the dyes was adjusted in order to obtain a loading of 10.0 μmol g-1. UV-vis spectra
of the solutions after dye adsorption showed that the amount of residual dyes is negligible,
confirming the quantitative adsorption of the dyes on Pt/TiO2.
5.3.3 Photocatalytic H2 Production Experiments
The sensitized Pt/TiO2 photocatalysts prepared were tested for H2 production under Vis
7.0. The experiments were performed adopting the same conditions previously optimized for PTZ-
based photocatalysts (262). No H2 production was observed using the bare Pt/TiO2 under the same
experimental conditions. Measured H2 production rates and H2 productivity versus irradiation time
are presented in Figure 5.15 and 5.16, respectively. TON and LFE20 calculated after 20 h of
irradiation are presented in Figure 5.17 (obtained values are listed in Table 5.5).
All of the investigated catalysts showed remarkable stability over a reasonable irradiation
time of 20 h (Figure 5.15). CBZ-sensitized photocatalysts showed by far the highest H2 productivity,
TON and LFE values. Namely, performances were at least one order of magnitude higher than
191
those referred to the benchmark PTZ-Th dye. Amongst CBZ based dyes, the photocatalytic activity
of CBZ-Th was considerably higher than that of CBZ-Fu. The same relative trend was recorded for
the PTZ family. Both POZs-sensitized photocatalysts demonstrated very small activity in H2
production.
Figure 5.15
H2 production rates from TEOA 10% v/v solution at pH = 7.0 under
irradiation with visible light ( > 420 nm) over Pt/TiO2 materials
sensitized with PTZs, POZs and CBZs dyes.
Figure 5.16
H2 production from TEOA 10% v/v solution at pH = 7.0 under
irradiation with visible light ( > 420 nm) over Pt/TiO2 materials
sensitized with PTZs, POZs and CBZs dyes.
192
Figure 5.17
Turnover number (TON) and Light-to-Fuel Efficiencies (LFE) calculated
from H2 production using TEOA/HCl solution at pH = 7.0 under
irradiation with visible light ( > 420 nm) over Pt/TiO2 materials
sensitized with PTZs, POZs and CBZs dyes.
Table 5.5 TON values and LFE20 for PTZs, POZs, and CBZs sensitized catalysts.
193
The extraordinary high activity of CBZ-sensitized systems may be explained by the visible
light-induced charge transfer method according to which they operate. Two methods of charge
transfer are known in the literature: dye-sensitization and ligand-to-metal charge transfer (LMCT)
(278, 279). In the first case, the dye molecules located at the TiO2/solution interface are
photoexcited and subsequently inject electrons from their LUMO* into the TiO2 CB. On the other
hand, the LMCT mechanism is obtained if charge transfer takes place after the formation of a
complex between TiO2 and the adsorbed dye. In this case there is a strong coupling between the
molecular orbital (HOMO) of the adsorbate and the energy band of the semiconductor, so that
excited electrons go directly from the ground state of the adsorbate to the semiconductor CB,
without involving the excited state of the adsorbate. It was shown in the literature that CBZ
derivatives (with a molecular structure very similar to that of CBZ-Fu and CBZ-Th) give a red-
shifted visible light absorption (up to 600 nm) when adsorbed on the surface of TiO2, typical of the
sensitization by LMCT (269). Such improved visible light absorption was not reported for PTZ and
POZ molecules (280), suggesting that these molecules operate mainly through the dye-
sensitization pathway.
Preliminary results of photocatalytic H2 generation using a cut-off filter at 515 nm (Figure
5.18) suggest that CBZ-based systems may operate following a different mechanism than PTZ and
POZ-based ones. In these experiments, a cut-off filter at 515 nm was added after activation of the
photocatalysts for 8 h under usual conditions. Under irradiation using photons, CBZ-Th-
based systems showed an appreciable photocatalytic activity, even if free CBZ in solution does not
absorb photons having (Figure 5.13). This result is an indication a new electron transition,
based on LMCT, should be involved in the photoexcitation of the system at such wavelengths. This
hypothesis will be tested by Diffused Reflectance Spectroscopy (DRS) experiments on the
dye/Pt/TiO2 systems, which will allow to observe the final system UV-vis spectra.
On the other hand, PTZ-Th sensitized photocatalysts were completely inactive, despite UV-
vis spectra of PTZ compounds show an intense light absorption up to 600 nm, due to ICT. This
observation suggests that the ICT band, responsible for HOMO-LUMO transition and dye-
sensitization of Pt/TiO2, is not efficient in promoting the photocatalytic H2 production (at least in
the part at higher wavelength). Taking this into account, the activity observed under irradiation
with > 420 nm using PTZ and POZ dyes as sensitizers (Figure 5.15-5.17) could be related to
absorption of light by - * band. This would also explain the very low activity observed using
POZ sensitizers, since electrons in POZ - cannot be excited by visible light.
194
Figure 5.18
H2 production rates from TEOA 10% v/v solution at pH = 7.0 over
Pt/TiO2 materials sensitized with PTZ-Th and CBZ-Th dyes: after
activation under irradiation with visible light ( > 420 nm) for 8h, the
photocatalytic activity under irradiation with photons with > 515 nm
is presented.
5.3.4 Conclusions
In conclusion, a rational design of the sensitizers was shown to be the key to obtain greatly
enhanced performances in H2 photogeneration under visible light irradiation. The presence of
heteroatoms in the donor group and π-spacer moieties of the investigated sensitizers dramatically
affected the photocatalytic H2 production in the visible range over Pt/TiO2. In particular, the
combination of planar donor cores (CBZ) with thiophene heterocyclic spacers (Th) afforded state-
of-the-art efficiency for dye-sensitized H2 generation.
195
5.4 Future Perspectives for Sustainable H2 Production
The sustainability of a process is a complex property to estimate: it encompasses
economical, environmental and social aspects of every step involved in the production of the final
commodity, and the metrics for its measurements are many, contextual, and still evolving. In
chemistry, a series of sustainability principles and guidelines were established in the 90's to guide
the practice of so-called green chemistry, in the context of increasing attention to problems of
chemical pollution and resource depletion. Photocatalysis meets many of the green chemistry
practical guidelines, and is surely one of the most green technologies existing for energy
production. Photocatalytic systems operate at room temperature, in water solutions, utilizing
clean, renewable solar light as the driving force, and cheap, nontoxic, chemically and biologically
inert TiO2-based materials as photocatalysts. Nonetheless, the sustainability of photocatalytic
processes, such as H2 production over sensitized photocatalysts, may be further improved.
The factors limiting the sustainability of the photocatalytic H2 production process studied
in the previous sections are the utilization of Pt noble metal as co-catalyst, of complex organic dyes
as sensitizers (even if in very little amounts) and of corrosive and irritating TEOA as a sacrificial
agent. In the context of this work, Pt/TiO2 systems were chosen as a benchmark in order to study
the performance of a series of sensitizers and rationalize the photocatalytic activity and stability in
terms of their structure. However, more sustainable alternatives to Pt exist (e.g. Pd (281), Ni (282),
Cu (283)) and the amount of co-catalyst can be diminished to 0.1 wt%, retaining photocatalytic
activity (282).
The intrinsic impact of organic sensitizers on the process sustainability is dependent on the
dye loading in the final catalyst (typically 0.5-1 wt% herein), the efficiency and modality of the
staining process, the dye stability against decomposition and leaching over irradiation time, dye
toxicity and synthesis procedure. In this work, the dye loading was optimized (see Section 5.2) and
a simple staining process was followed to prepare efficient and stable sensitized photocatalysts.
The optimization of the dye synthetic procedures and an evaluation of their impact on the
environment and human health are out of the scope of this thesis. However, the synthetic
procedures employed by the group of Professor Alessandro Abbotto comprise only a few steps and
have been optimized keeping in mind the principles of green chemistry. It is also worth noting
that, being the photocatalyst stable, the process sustainability is affected much more by
(continuative) operational conditions than it is by (ideally, one-time) catalyst preparation.
TEOA was chosen as a sacrificial agent because it is a consolidated benchmark for
sensitized photocatalytic H2 production studies (46). However, for real applications, 10 v/v solutions
of TEOA would not be a very green choice, since the industrial synthesis of TEOA requires ethylene
196
oxide, a widely used chemical in view of its reactivity, which is also very harmful and hazardous on
the downside (284). A much greener class of sacrificial agents is that of oxygenated compounds.
Sugars, glycerol (side product of biodiesel production) and alcohols such as ethanol (EtOH ─ from
sugars fermentation) and methanol (MeOH ─ from glycerol steam reforming) can all be obtained by
biomass upgrading, and were demonstrated to work as sacrificial agents in photocatalytic H2
production (282, 285). However, to the best of our knowledge, except for one study of methanol
photoreforming (286), none of these compounds was studied in dye-sensitized photocatalytic H2
production.
Figure 5.20 shows preliminary results of photocatalytic H2 production using TEOA or EtOH
as a sacrificial agent, for Pt/TiO2 sensitized with a series of triphenilamine-based dyes containing
the spacer unit thiazolo[5,4-d]thiazole (TTZ), prepared in Dr. G. Reginato research group (ICCOM-
CNR, Florence), and previously studied in DSSC (287), namely TTZ3, TTZ4 and TTZ5 (Figure 5.19)10.
The systems were active with both TEOA and EtOH, and H2 production was very stable over
irradiation time. In TEOA solutions, the Pt/TiO2 sensitized with TTZ dyes performed slightly better
than PTZ-based systems (see Figure 5.15-5.16), with TTZ5 showing an activity almost double as
those of TTZ3 and TTZ4. When EtOH was used as sacrificial agent, the rates of H2 production were
lower, and interestingly, the activity trend was opposite than with TEOA, TTZ3>TTZ4>TTZ5.
Notably, when tested for H2 production with EtOH as a sacrificial agent, the previously reported
CBZs and PTZs-based dyes showed very modest activity (about the detection limit for the chosen
conditions) or no activity at all.
Figure 5.19 Molecular structure of TTZ3, TTZ4 and TTZ5.
10 For details on dyes synthesis and optical/redox properties, see reference (287)
197
Figure 5.20
H2 production rates (A,C) and amount (B,D), from TEOA 10% v/v
solution at pH = 7.0 (up) or EtOH 10% v/v solution, under irradiation
with visible light ( > 420 nm) over Pt/TiO2 materials sensitized with
TTZ3 (blue), TTZ4 (black) and TTZ5 (red).
The differences observed between TEOA and EtOH can be due to a number of factors, such
as different redox and adsorption properties of the sacrificial agents. A detailed study on EtOH
photoreforming by dye-sensitized photocatalysts and the mechanism of operation is under way.
198
6. Conclusions
In conclusion, the present work provides new insights into some important environmental
and energy-related catalytic reactions, taking advantage of well-defined nanostructured materials
and of state-of-the-art spectroscopic and imaging techniques in order to rationalize catalytic
performances. Three catalytic reactions were studied: methane catalytic oxidation, catalytic
hydrodeoxygenation of biomass-derived compounds and photocatalytic hydrogen production.
Regarding methane catalytic oxidation, nanostructured Pd@MOx-based catalysts were
synthesized by slightly modifying a self-assembly method (92), and their catalytic activity was
evaluated in conditions approaching the ones of industrial effluent gases and combustion engines
exhausts. In particular, the effect of the presence of H2O, phosphates or SO2 in the reaction mixture
was studied. Model catalysts based on Pd@MOx units were also developed in order to carry out
advanced photoelectron spectroscopy surface studies, essential to understand the deactivation
processes observed in the presence of the above mentioned poisoning compounds.
Hierarchical Pd@CeO2/Si-Al2O3 catalysts - otherwise stable under ideal conditions - are
progressively deactivated in the presence of water (Section 3.3). While at temperatures below
500 °C deactivation is easily reversed by removing the water, at higher temperatures an irreversible
deactivation process is observed. This is attributable to lower active phase accessibility, PdO
decomposition to Pd, and to the formation of stable cerium hydroxides, which are decomposed
only by high-temperature treatment. Lowering the amount of CeO2 in hierarchical catalysts leads
to enhanced stability thanks to the retention of more accessible active phase. The results show that
design of catalysts for methane oxidation must maximize metal-support interactions to favor
oxygen transfer from a reducible promoter, while keeping the active phase accessible to gas phase
reactants. Indeed, Pd@ZrO2 showed much greater hydrothermal stability (115).
The effect of phosphorus poisoning on the catalytic oxidation of methane over model
Pd@CeO2/graphite catalysts was found to be dramatically influenced by temperature and presence
of H2O (Section 3.4). While P-free Pd@CeO2/graphite catalysts were active and stable under all
studied conditions (500-600 °C; dry and wet conditions), P-poisoned catalysts were less active and
stable because of partial thermal sintering, worsening at higher temperatures and in the presence
of water vapor. The combination of XPS/SRPES, operando XANES measurements, SEM/EDS and
AFM techniques provided evidence of a temperature dependent, water-driven P-poisoning of Pd-
199
CeO2-based oxidation catalysts, proceeding via severe aggregation of ceria nanoparticles,
incorporation of Pd active phase in the bulk of the crystallites and exposure of CePO4 to the catalyst
surface.
Nanostructured Pd@CexZr1-xO2 (Pd@MOx) units were synthesized in the whole
compositional range (0<x<1) (Section 3.5). The synthesis of dispersed Pd@MOx allowed the
preparation of a series of high-surface-area Si-Al2O3 supported catalysts and model catalysts having
similar nanostructure and surface chemistry. Comparison of results on the two types of catalysts
allowed the SO2 poisoning of methane oxidation on Pd-based catalysts to be systematically studied
to elucidate the role of the MOx promoter and the aging conditions. At lower temperatures (<450
°C), the PdO active phase is irreversibly poisoned by SO2 due to interaction with sulfates which are
not able to spillover to the support/promoter. At higher temperatures (>500 °C), poisoning is
slowed by formation of sulfate species on the oxide promoter. Due to partial decomposition of
sulfates at 500 °C, Pd@ZrO2-based catalysts showed the best sulfur-poisoning resistance, attaining
complete regeneration even after prolonged aging, and thus they are the best candidates for real
application. [email protected] catalysts showed intermediate sulfur tolerance compared to Pd@CeO2
and Pd@ZrO2, in agreement with previously reported results (142). The high chemical sensitivity of
PES techniques provided direct evidence for previously suggested formation of sulfate species on
individual metal cations in CexZr1-xO2 mixed oxides (142). Finally, we proved that the model-catalyst
approaches developed here allow the study of metal-support interactions in other catalytically
relevant systems by simply varying the ALD-deposited thin film composition.
The present study suggests that new technological breakthroughs are needed in order to
solve the methane emission issue in vehicles. The hierarchical catalysts investigated herein are an
important step in the process of seeking a solution, but, despite their enhanced activity and
thermal stability, they suffer from deactivation in a similar way to traditional Pd-based catalysts.
Given the severity of the environment of vehicles exhausts and the low reactivity of methane, an
effective solution for methane abatement will require not only thermally stable catalysts but also
improvement in fuel quality and in engine and after-treatment engineering (e.g. electronically
controlled fuel injection to reduce unburned HCs emissions).
Regarding HDO catalytic reaction studies, we first demonstrated that the products
distribution for furfural HDO is greatly influenced by H2 pressure (Section 4.2). While at lower H2
pressures decarbonylation is a major reaction pathway, higher H2 pressures strongly favor the HDO
reaction. When studied under the same conditions, HDO of furfural and HMF are very similar, both
following a sequential reaction network, with the desired products (MF or DMF) formed as
200
intermediates. The development of unifying concepts for different reactants and reaction
conditions are important for understanding these reactions.
In a series of studies (Sections 4.3-4.5), we investigated the performances of well-defined
metal nanocrystalline alloys in the liquid-phase HDO reaction of HMF in order to understand the
factors affecting the catalysts selectivity and stability. Pt-based (Pt-Co, Pt-Zn, Pt-Cu, Pt-Ni) and Ni-Cu
nanocrystalline alloys of controlled size and compositions were synthesized by advanced
solvothermal methods and supported on high-surface area carbon. Thanks to the uniformity of the
synthesized materials, the performances observed in HDO reactions were rationalized on the basis
of nanocrystals composition. Alloying was shown to have a positive effect on catalyst selectivity
and stability, and optimal compositions were found for each alloy. The critical factor for achieving
high selectivity to DMF in the HDO reaction of HMF appears to be the bonding mode of DMF,
which has to prevent furan ring from lying down on the surface and be activated. However, there
may be multiple ways to achieve such preferential bonding.
In the case of Pt-Co nanocrystal-based catalyst, an optimal DMF yield (98%) was obtained
over Pt3Co2/C, for which a Co oxide monolayer formed on the surface of the metallic core. The thin
overlayer induces high DMF yields because it weakly interacts with the furan ring, preventing over-
hydrogenation and ring opening of DMF to secondary by-products while forming active sites to
carry out the HDO process. In this regard, composition control is crucial to cover the entire surface
with an oxide layer and avoid exposed metallic patches that could promote side reactions (185).
However, this oxide overlayer model is not expected to be suitable for Pt-Ni, Pt-Zn, and Pt-Cu
alloys, not as oxophilic as Co but just as selective. Indeed, the results of the HDO study over Ni-Cu
base-meal catalysts strongly indicate that an oxide overlayer is not necessary to achieve high DMF
yield in the HDO reaction of HMF.
The fact that high DMF yields were observed over Ni-Cu alloys catalysts demonstrates that
reasonably inexpensive materials can be used for selective HDO of HMF. It is interesting to ask
whether these base-metal alloys could also be used for other HDO reactions, such as would be
important in the processing of lignin. There are indeed indications for this (244).
Regarding H2 photocatalytic production experiments, benchmark Pt/TiO2 photocatalysts
were sensitized with a series of donor-acceptor organic dyes. A rational design of the sensitizers
was shown to be the key to obtain greatly enhanced performances in H2 photocatalytic production
under visible light irradiation. We investigated the effect of ancillary substituents hydrophilicity, of
the D-core nature (PTZ, CBZ, POZ) and of heteroatoms substitution in the spacer units. The
combination of planar donor cores (CBZ) with thiophene heterocyclic spacers (Th) afforded state-
of-the-art efficiency for dye-sensitized H2 generation using benchmark TEOA as sacrificial agent.
201
Moreover, the general and scalable synthetic approach can be adapted to a large variety of
derivatives, to synthesize photosensitizers having finely tuned properties, and to study their effect
on the efficiency of H2 photocatalytic production.
Finally, stable and sizeable H2 production was observed over TTZ-sensitized Pt/TiO2
photocatalysts using EtOH as a sacrificial agent, which is a very promising result for the
development of photocatalytic systems of improved sustainability.
202
7. Acknowledgments
Voglio ringraziare il Professor Paolo Fornasiero per essere stato in questi anni non solo un
grande mentore scientifico, ma anche un esempio di vita e un amico. Grazie per quel dollaro
portafortuna, che ancora porto con me. Avrà forse poco valore sulla carta, ma è un simbolo
prezioso, delle possibilità che mi hai offerto e dei momenti divertenti passati assieme.
A tutti i componenti del gruppo di ricerca di Fornasiero, grazie. Grazie Tiziano, Valentina,
Alessandro, Michele, Marta e Anne varie ed eventuali. Grazie anche al Professor Mauro Graziani, per
i preziosi consigli e gli aneddoti, scientifici e non. Un giorno, quando le querce di Paolo saranno alte
e non andranno più potate ogni inverno, spero di sedermi lì nel suo campo con voi tutti e ricordare
questi anni magnifici.
In particolare, ringrazio il Dott. Tiziano Montini per avermi insegnato a fare un sacco di cose
che un bravo ricercatore dovrebbe fare. Ad esempio, a farmi le domande giuste. Ad essere
tempestivo, ma anche a badare ai dettagli. A trattare gli strumenti con cura, e ad aprirli
all'occorrenza. Per tutte le avventure passate assieme, da amico, ti ringrazio di cuore.
I would like to thank Professor Ray Gorte, for the great and valuable time I spent in his lab.
It's been a great honor for me to work there. It was very inspiring to see him step into the lab every
morning to discuss with each of his students, being always very smart and kind. I will also never
forget my trip to Baltimore, which I visited thanks to his advice.
I'm also grateful to Professor Chris Murray for giving me the opportunity to work in his
awesome lab. It has been a great professional and human experience.
Grazie a Emiliano Fonda e ai suoi colleghi della linea Samba al sincrotrone di Soleil. Con la
sua simpatia, pazienza e perseveranza Emiliano riesce a rendere leggere le più lunghe giornate di
beamtime, quando ormai ogni cognizione del giorno e della notte è andata persa.
Oggi come tre anni fa voglio ringraziare tutti i chimici del C11, passati, presenti e futuri. Fra
di loro, un pensiero affettuoso è rivolto a Massimo (Pessimo) Rigo, Giulio Ragazzon (Picio!),
Margherita Macino (tha-tha-tha Morgue!) e L.E.M.S. (Lukas Eugen Marsoner Steinkasserer).
Grazie alla mia famiglia, per il sostegno e l'aiuto insostituibile.
203
Grazie ai miei amici, vicini e lontani. Grazie a Seba, Nick, Braida e Pier. Thank you Cong, for
being a great friend in a stranger land - we shall go to another rock concert one day! Thanks to
Ming, Jing, Jennifer, and all the guys from Gorte's lab and Murray's lab.
and everyone in Matolin's group.
Grazie, Serena (RNH2).
A tutti voi, grazie.
204
8. Bibliography
(1) Monastersky, R. Anthropocene: The human age. Nature 2015, 519, 144 147.
(2) Lazarus, R. J. The Making of Environmental Law; University of Chicago Press, 2004.
(3) Whitesides, G. M. Reinventing Chemistry. Angew. Chemie Int. Ed. 2015, 54, 3196 3209.
(4) Ham, D. J.; Phuruangrat, A.; Thongtem, S.; Lee, J. S. Hydrothermal synthesis of monoclinic WO3 nanoplates and nanorods used as an electrocatalyst for hydrogen evolution reactions from water. Chem. Eng. J. 2010, 165, 365 369.
(5) Simon, Q.; Barreca, D.; Gasparotto, A.; Maccato, C.; Montini, T.; Gombac, V.; Fornasiero, P.; Lebedev, O. I.; Turner, S.; Van Tendeloo, G. Vertically oriented CuO/ZnO nanorod arrays: from plasma-assisted synthesis to photocatalytic H2 production. J. Mater. Chem. 2012, 22, 11739.
(6) Li, S.; Liu, H.; Zhao, X.; Wang, X. A cost-effective preparation method of Ba-hexaaluminate nanoparticles for catalytic combustion of methane. Mater. Lett. 2007, 61, 3935 3938.
(7) Viñes, F.; Lykhach, Y.; Staudt, T.; Lorenz, M. P.; Papp, C.; Steinrück, H.-P.; Libuda, J.; Neyman, K. M.; Görling, A. Methane activation by platinum: critical role of edge and corner sites of metal nanoparticles. Chemistry 2010, 16, 6530 9.
(8) Cargnello, M.; Jaen, J. J. D.; Garrido, J. C. H.; Bakhmutsky, K.; Montini, T.; Gamez, J. J. C.; Gorte, R. J.; Fornasiero, P. Exceptional Activity for Methane Combustion over Modular Pd@CeO2 Subunits on Functionalized Al2O3. Science 2012, 337, 713 717.
(9) Heck, R.; Farrauto, R. Automobile exhaust catalysts. Appl. Catal. A Gen. 2001, 221, 443 457.
(10) Skoglundh, M.; Fridell, E. Strategies for enhancing low-temperature activity. Top. Catal. 2004, 28, 7987.
(11) EEA ─ European Environment Agency. Air quality in Europe -2011 report; 2011.
(12) EEA ─ European Environment Agency. The contribution of transport to air quality; 2012.
(13) Seinfeld, J.; Pandis, S. Atmospheric chemistry and physics: from air pollution to climate change; 2nd ed.; Wiley, 2006.
(14) Lelieveld, J.; Crutzen, P. J.; Dentener, F. J. Changing concentration, lifetime and climate forcing of atmospheric methane. Tellus B 1998, 50, 128 150.
(15) Turner, A. J.; Jacob, D. J.; Benmergui, J.; Wofsy, S. C.; Maasakkers, J. D.; Butz, A.; Hasekamp, O.; Biraud, S. C. A large increase in U.S. methane emissions over the past decade inferred from satellite data and
surface observations. Geophys. Res. Lett. 2016, 43, 2218 2224.
(16) Everaert, K.; Baeyens, J. Catalytic combustion of volatile organic compounds. J. Hazard. Mater. 2004, 109, 113 39.
(17) Li, W. B.; Wang, J. X.; Gong, H. Catalytic combustion of VOCs on non-noble metal catalysts. Catal. Today 2009, 148, 81 87.
(18) Khan, F. I.; Kr. Ghoshal, A. Removal of Volatile Organic Compounds from polluted air. J. Loss Prev. Process Ind. 2000, 13, 527 545.
205
(19) Spivey, J. Complete catalytic oxidation of volatile organics. Ind. Eng. Chem. Res. 1987, 2165 2180.
(20) Wang, J.; Jin, L.; Gao, J.; Shi, J.; Zhao, Y.; Liu, S.; Jin, T.; Bai, Z.; Wu, C.-Y. Investigation of speciated VOC in
gasoline vehicular exhaust under ECE and EUDC test cycles. Sci. Total Environ. 2013, 445 446, 110 6.
(21) Gélin, P.; Primet, M. Complete oxidation of methane at low temperature over noble metal based catalysts: a review. Appl. Catal. B Environ. 2002, 39, 1 37.
(22) Yin, J.; Su, S.; Yu, X. X.; Weng, Y. Thermodynamic characteristics of a low concentration methane catalytic combustion gas turbine. Appl. Energy 2010, 87, 2102 2108.
(23) Lampert, J.; Kazi, M.; Farrauto, R. Palladium catalyst performance for methane emissions abatement from lean burn natural gas vehicles. Appl. Catal. B Environ. 1997, 14, 211 223.
(24) Armstrong, R. C.; Wolfram, C.; de Jong, K. P.; Gross, R.; Lewis, N. S.; Boardman, B.; Ragauskas, A. J.;
Ehrhardt-Martinez, K.; Crabtree, G.; Ramana, M. V. The frontiers of energy. Nat. Energy 2016, 1, 15020.
(25) Ethanol Benefits and Considerations; 2014.
(26) EPA Lifecycle Analysis of Greenhouse Gas Emissions from Renewable Fuels; 2010.
(27) EASAC. The current status of biofuels in the European Union, their environmental impacts and future prospects; 2012.
(28) Fu, C.; Mielenz, J. R.; Xiao, X.; Ge, Y.; Hamilton, C. Y.; Rodriguez, M.; Chen, F.; Foston, M.; Ragauskas, A.; Bouton, J.; Dixon, R. A.; Wang, Z.-Y. Genetic manipulation of lignin reduces recalcitrance and improves ethanol production from switchgrass. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 3803 3808.
(29) Furimsky, E. Catalytic hydrodeoxygenation. Appl. Catal. A Gen. 2000, 199, 147 190.
(30) Luo, J.; Monai, M.; Yun, H.; Arroyo-Ramírez, L.; Wang, C.; Murray, C. B.; Fornasiero, P.; Gorte, R. J. The H2 Pressure Dependence of Hydrodeoxygenation Selectivities for Furfural over Pt/C Catalysts. Catal. Letters 2016, 146, 711 717.
(31) Renew. Sustain. Energy Rev. 2011, 15, 3034 3040.
(32) Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Dye-Sensitized Solar Cells. Chem. Rev. 2010, 110, 6595 6663.
(33) Albers, P.; Pietsch, J.; Parker, S. F. Poisoning and deactivation of palladium catalysts. J. Mol. Catal. A
Chem. 2001, 173, 275 286.
(34) Larese, C.; López Granados, M.; Mariscal, R.; Fierro, J. L. G.; Lambrou, P. S.; Efstathiou, A. M. The effect of calcination temperature on the oxygen storage and release properties of CeO2 and Ce Zr O metal oxides modified by phosphorus incorporation. Appl. Catal. B Environ. 2005, 59, 13 25.
(35) Lopez Granados, M.; Galisteo, F.; Lambrou, P.; Mariscal, R.; Sanz, J.; Sobrados, I.; Fierro, J.; Efstathiou, A. Role of P-containing species in phosphated CeO2 in the deterioration of its oxygen storage and release properties. J. Catal. 2006, 239, 410 421.
(36) Christou, S. Y.; Álvarez-Galván, M. C.; Fierro, J. L. G.; Efstathiou, A. M. Suppression of the oxygen storage and release kinetics in Ce0.5Zr0.5O2 induced by P, Ca and Zn chemical poisoning. Appl. Catal. B Environ. 2011, 106, 103 113.
(37) . . Cerium phosphate in automotive exhaust catalyst poisoning. Appl. Catal. B Environ. 2004, 50, 113 125.
(38) Sitthisa, S.; An, W.; Resasco, D. E. Selective conversion of furfural to methylfuran over silica-supported NiFe bimetallic catalysts. J. Catal. 2011, 284, 90 101.
(39) Cargnello, M.; Doan-Nguyen, V. V. T.; Gordon, T. R.; Diaz, R. E.; Stach, E. A.; Gorte, R. J.; Fornasiero, P.;
206
Murray, C. B. Control of Metal Nanocrystal Size Reveals Metal-Support Interface Role for Ceria Catalysts. Science 2013, 341, 771 773.
(40) Cargnello, M.; Chen, C.; Diroll, B. T.; Doan-Nguyen, V. V. T.; Gorte, R. J.; Murray, C. B. Efficient Removal of Organic Ligands From Supported Nanocrystals by Fast Thermal Annealing Enables Catalytic Studies on Well Defined Active Phases. J. Am. Chem. Soc. 2015, 137, 6906 6911.
(41) Luo, J.; Arroyo-Ramírez, L.; Gorte, R. J.; Tzoulaki, D.; Vlachos, D. G. Hydrodeoxygenation of HMF over
Pt/C in a continuous flow reactor. AIChE J. 2015, 61, 590 597.
(42) Schneider, J.; Matsuoka, M.; Takeuchi, M.; Zhang, J.; Horiuchi, Y.; Anpo, M.; Bahnemann, D. W. Understanding TiO2 Photocatalysis: Mechanisms and Materials. Chem. Rev. 2014, 114, 9919 9986.
(43) Liu, X.; Long, J.; Wang, G.; Pei, Y.; Zhao, B.; Tan, S. Effect of structural modification on the performances of phenothiazine-dye sensitized solar cells. Dye. Pigment. 2015, 121, 118 127.
(44) Sathiyan, G.; Sivakumar, E. K. T.; Ganesamoorthy, R.; Thangamuthu, R.; Sakthivel, P. Review of carbazole based conjugated molecules for highly efficient organic solar cell application. Tetrahedron Lett. 2016, 57, 243 252.
(45) Kisch, H.; Bahnemann, D. Best Practice in Photocatalysis: Comparing Rates or Apparent Quantum Yields? J. Phys. Chem. Lett. 2015, 6, 1907 1910.
(46) Cecconi, B.; Manfredi, N.; Montini, T.; Fornasiero, P.; Abbotto, A. Dye-Sensitized Solar Hydrogen
Production: The Emerging Role of Metal-Free Organic Sensitizers. European J. Org. Chem. 2016, 2016, 5194 5215.
(47) Thomas, J. M. and W. J. Principle and Practice of Heterogeneous Catalysis; VCH Publisher: Weinheim (Germany), 1997.
(48) Cheng, H.; Wang, J.; Zhao, Y.; Han, X. Effect of phase composition, morphology, and specific surface area on the photocatalytic activity of TiO2 nanomaterials. RSC Adv. 2014, 4, 47031 47038.
(49) Cozzoli, P. D.; Kornowski, A.; Weller, H. Low-Temperature Synthesis of Soluble and Processable Organic-Capped Anatase TiO2 Nanorods. J. Am. Chem. Soc. 2003, 125, 14539 14548.
(50) Pierotti, R. A.; Rouquerol, J.; Sing, K. S. W.; Everett, D. H.; Haul, R. A. W.; Moscou, L.; Siemieniewska, T. Reporting physisorption data for gas/solid systems with special reference to the determination of surface area and porosity. Pure Appl. Chem. 1985, 57, 603 619.
(51) Rouquerol, J.; Avnir, D.; Fairbridge, C. W.; Everett, D. H.; Haynes, J. M.; Pernicone, N.; Ramsay, J. D. F.; Sing, K. S. W.; Unger, K. K. Recommendations for the characterization of porous solids. Pure Appl. Chem. 1994, 66, 1739 1758.
(52) Barrett, E.; Joyner, L.; Halenda, P. The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms. J. Am. Chem. Soc. 1951, 73, 373 380.
(53) Dulaurent, O.; Bianchi, D. Adsorption isobars for CO on a Pt/Al2O3 catalyst at high temperatures using FTIR spectroscopy : isosteric heat of adsorption and adsorption model. Appl. Catal. A Gen. 2000, 196, 271 280.
(54) Gatica, J.; Baker, R.; Fornasiero, P.; Bernal, S.; Kaspar, J. Characterization of the metal phase in NM/Ce0.68Zr0.32O2 (NM: Pt and Pd) catalysts by hydrogen chemisorption and HRTEM microscopy: A
comparative study. J. Phys. Chem. B 2001, 2, 1191 1199.
(55) Zhang, J.; Wang, Z.; Wang, Z. X. Adsorption of carbon monoxide on Pd low-index surfaces and (110) reconstructed surface. Surf. Interface Anal. 2011, 43, 1038 1045.
(56) Ramos, I. A. C.; Montini, T.; Lorenzut, B.; Troiani, H.; Gennari, F. C.; Graziani, M.; Fornasiero, P. Hydrogen production from ethanol steam reforming on M/CeO2/YSZ (M=Ru, Pd, Ag) nanocomposites. Catal. Today 2012, 180, 96 104.
207
(57) Saghi, Z.; Midgley, P. A. Electron Tomography in the (S)TEM: From Nanoscale Morphological Analysis to 3D Atomic Imaging. Annu. Rev. Mater. Res. 2012, 42, 59 79.
(58) Goodhew, P. J.; Humphreys, J.; Beanland, R. Electron Microscopy and Analysis; Taylor & Francis, 2001; Vol. 54.
(59) Yang, J. C.; Small, M. W.; Grieshaber, R. V; Nuzzo, R. G. Recent developments and applications of electron microscopy to heterogeneous catalysis. Chem. Soc. Rev. 2012, 41, 8179 8194.
(60) Koningsberger, D. C.; Prins, R. X-Ray Absorption: Principles, Applications, Techniques of EXAFS, SEXAFS and XANES; 1987.
(61) Als-Nielsen, J.; McMorrow, D. Elements of Modern X-ray Physics, 2nd Edition; 2011.
(62) Bunker, G. Introduction to XAFS; 2010.
(63) La Fontaine, C.; Barthe, L.; Rochet, A.; Briois, V. X-ray absorption spectroscopy and heterogeneous Catal. Today 2013, 205, 148 158.
(64) Ankudinov, A.; Ravel, B.; Rehr, J.; Conradson, S. Real-space multiple-scattering calculation and interpretation of X-ray-absorption near-edge structure. Phys. Rev. B 1998, 58, 7565 7576.
(65) Ravel, B.; Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray absorption spectroscopy using IFEFFIT. J. Synchrotron Radiat. 2005, 12, 537 541.
(66) van der Heide, P. X-Ray Photoelectron Spectroscopy; John Wiley & Sons, Inc.: Hoboken, NJ, USA, 2011.
(67) Barber, J. Photosynthetic energy conversion: natural and artificial. Chem. Soc. Rev. 2009, 38, 185 196.
(68) Hurtado, P.; Ordóñez, S.; Vega, A.; Díez, F. V. Catalytic combustion of methane over commercial
catalysts in presence of ammonia and hydrogen sulphide. Chemosphere 2004, 55, 681 9.
(69) Colussi, S.; De Leitenburg, C.; Dolcetti, G.; Trovarelli, A. The role of rare earth oxides as promoters and stabilizers in combustion catalysts. J. Alloys Compd. 2004, 374, 387 392.
(70) Corro, G.; Cano, C.; Fierro, J. L. G. A study of Pt -Al2O3 catalysts for methane oxidation resistant to deactivation by sulfur poisoning. J. Mol. Catal. A Chem. 2010, 315, 35 42.
(71) Venezia, A. M.; Di Carlo, G.; Pantaleo, G.; Liotta, L. F.; Melaet, G.; Kruse, N. Oxidation of CH4 over Pd supported on TiO2-doped SiO2: Effect of Ti(IV) loading and influence of SO2. Appl. Catal. B Environ. 2009, 88, 430 437.
(72) Ciuparu, D.; Lyubovsky, M. R.; Altman, E.; Pfefferle, L. D.; Datye, A. Catalytic Combustion of Methane Over Palladium-Based Catalysts. Catal. Rev. 2002, 44, 593 649.
(73) Schwartz, W. R.; Pfefferle, L. D. Combustion of Methane over Palladium-Based Catalysts: Support
Interactions. J. Phys. Chem. C 2012, 116, 8571 8578.
(74) Chin, Y.-H. C.; Buda, C.; Neurock, M.; Iglesia, E. Consequences of metal-oxide interconversion for C-H bond activation during CH4 reactions on Pd catalysts. J. Am. Chem. Soc. 2013, 135, 15425 42.
(75) Colussi, S.; Trovarelli, A.; Groppi, G.; Llorca, J. The effect of CeO2 on the dynamics of Pd PdO transformation over Pd/Al2O3 combustion catalysts. Catal. Commun. 2007, 8, 1263 1266.
(76) Mayernick, A. D.; Janik, M. J. Methane oxidation on Pd Ceria: A DFT study of the mechanism over PdxCe O2, Pd, and PdO. J. Catal. 2011, 278, 16 25.
(77) Doornkamp, C.; Ponec, V. The universal character of the Mars and Van Krevelen mechanism. J. Mol.
Catal. A Chem. 2000, 162, 19 32.
(78) Mars, P.; van Krevelen, D. W. Oxidations carried out by means of vanadium oxide catalysts. Chem. Eng.
Sci. 1954, 3, 41 59.
208
(79) Sayle, T.; Cantoni, M.; Bhatta, U. Strain and Architecture-Tuned Reactivity in Ceria Nanostructures; Enhanced Catalytic Oxidation of CO to CO2. Chem. Mater. 2012, 24, 1811 1821.
(80) Rocchini, E.; Vicario, M.; Llorca, J.; de Leitenburg, C.; Dolcetti, G.; Trovarelli, A. Reduction and Oxygen Storage Behavior of Noble Metals Supported on Silica-Doped Ceria. J. Catal. 2002, 211, 407 421.
(81) Trovarelli, A. Catalytic Properties of Ceria. Catal. Rev. 2006, 38, 439 520.
(82) Zhou, G.; Shah, P. R.; Montini, T.; Fornasiero, P.; Gorte, R. J. Oxidation enthalpies for reduction of ceria surfaces. Surf. Sci. 2007, 601, 2512 2519.
(83) De Rogatis, L.; Cargnello, M.; Gombac, V.; Lorenzut, B.; Montini, T.; Fornasiero, P. Embedded phases: a way to active and stable catalysts. ChemSusChem 2010, 3, 24 42.
(84) Mowery, D. L.; McCormick, R. L. Deactivation of alumina supported and unsupported PdO methane
oxidation catalyst: the effect of water on sulfate poisoning. Appl. Catal. B Environ. 2001, 34, 287 297.
(85) Ciuparu, D.; Pfefferle, L. Support and water effects on palladium based methane combustion catalysts. Appl. Catal. A Gen. 2001, 209, 415 428.
(86) Ciuparu, D.; Katsikis, N.; Pfefferle, L. Temperature and time dependence of the water inhibition effect on supported palladium catalyst for methane combustion. Appl. Catal. A Gen. 2001, 216, 209 215.
(87) Persson, K.; Pfefferle, L. D.; Schwartz, W.; Ersson, A.; Järås, S. G. Stability of palladium-based catalysts during catalytic combustion of methane: The influence of water. Appl. Catal. B Environ. 2007, 74, 242250.
(88) Ciuparu, D.; Perkins, E.; Pfefferle, L. In situ DR-FTIR investigation of surface hydroxyls on γ-Al2O3 supported PdO catalysts during methane combustion. Appl. Catal. A Gen. 2004, 263, 145 153.
(89) Monai, M.; Montini, T.; Chen, C.; Fonda, E.; Gorte, R. J.; Fornasiero, P. Methane Catalytic Combustion over Hierarchical Pd@CeO2/Si-Al2O3: Effect of the Presence of Water. ChemCatChem 2015, 7, 20382046.
(90) Honkane, M.; Kärkkäinen, M.; Kolli, T.; Heikkinen, O.; Viitanen, V.; Zeng, L.; Jiang, H.; Kallinen, K.; Huuhtanen, M.; Keiski, R. L.; Lahtinen, J.; Olsson, E.; Vippola, M. Accelerated Deactivation Studies of the Natural-Aging. Appl. Catal. B Environ. 2015.
(91) Farrauto, R. J. Low-Temperature Oxidation of Methane. Science 2012, 337, 659 660.
(92) Bakhmutsky, K.; Wieder, N. L.; Cargnello, M.; Galloway, B.; Fornasiero, P.; Gorte, R. J. A versatile route to core-shell catalysts: synthesis of dispersible M@oxide (M=Pd, Pt; oxide=TiO2, ZrO2) nanostructures by self-assembly. ChemSusChem 2012, 5, 140 8.
(93) Cargnello, M.; Wieder, N. Synthesis of Dispersible Pd@CeO2 Core Shell Nanostructures by Self-Assembly. J. Am. Chem. Soc. 2010, 1402 1409.
(94) Gradeff, P.; Schreiber, F. A simplified method for the synthesis of ceric alkoxides from ceric ammonium nitrate. Inorg. Chem. 1985, 1110 1111.
(95) Adijanto, L.; Bennett, D. A.; Chen, C.; Yu, A. S.; Cargnello, M.; Fornasiero, P.; Gorte, R. J.; Vohs, J. M. Exceptional thermal stability of Pd@CeO2 core-shell catalyst nanostructures grafted onto an oxide
surface. Nano Lett. 2013, 13, 2252 2257.
(96) Papaefthimiou, P.; Ioannides, T.; Verykios, X. Catalytic incineration of volatile organic compounds present in industrial waste streams. Appl. Therm. Eng. 1998, 18, 1005 1012.
(97) Burch, R.; Urbano, F.; Loader, P. K. Methane combustion over palladium catalysts: The effect of carbon dioxide and water on activity. Appl. Catal. A Gen. 1995, 123, 173 184.
(98) Hurtado, P.; Ordonez, S.; Sastre, H.; Diez, F. V. Combustion of methane over palladium catalyst in the
209
presence of inorganic compounds: inhibition and deactivation phenomena. Appl. Catal. B Environ. 2004, 47, 85 93.
(99) Kikuchi, R.; Maeda, S.; Sasaki, K.; Wennerström, S.; Ozawa, Y.; Eguchi, K. Catalytic activity of oxide-supported Pd catalysts on a honeycomb for low-temperature methane oxidation. Appl. Catal. A Gen. 2003, 239, 169 179.
(100) Schwartz, W. R.; Ciuparu, D.; Pfefferle, L. D. Combustion of Methane over Palladium-Based Catalysts: Catalytic Deactivation and Role of the Support. J. Phys. Chem. C 2012, 116, 8587 8593.
(101) Datye, A. K.; Bravo, J.; Nelson, T. R.; Atanasova, P.; Lyubovsky, M.; Pfefferle, L. Catalyst microstructure and methane oxidation reactivity during the Pd↔PdO transformation on alumina supports. Appl. Catal. A Gen. 2000, 198, 179 196.
(102) Zhang, F.; Wang, P.; Koberstein, J.; Khalid, S.; Chan, S.-W. S. Cerium oxidation state in ceria nanoparticles studied with X-ray photoelectron spectroscopy and absorption near edge spectroscopy. Surf. Sci. 2004, 563, 10726 10732.
(103) Keating, J.; Sankar, G.; Hyde, T. I.; Kohara, S.; Ohara, K. Elucidation of structure and nature of the PdO-Pd transformation using in situ PDF and XAS techniques. Phys. Chem. Chem. Phys. 2013, 15, 8555 65.
(104) Badri, A.; Binet, C.; Lavalley, J. An FTIR study of surface ceria hydroxy groups during a redox process with H2. J. Chem. Soc., Faraday Trans. 1996, 92, 4669 4673.
(105) Setiawan, A.; Friggieri, J.; Kennedy, E. M.; Dlugogorski, B. Z.; Stockenhuber, M. Catalytic combustion of ventilation air methane (VAM) - long term catalyst stability in the presence of water vapour and mine dust. Catal. Sci. Technol. 2014, 4, 1793 1802.
(106) Yang, S.; Maroto-Valiente, A.; Benito-Gonzalez, M.; Rodriguez-Ramos, I. Methane combustion over supported palladium catalysts I . Reactivity and active phase. Appl. Catal. B Environ. 2000, 28, 223 233.
(107) Esch, F.; Fabris, S.; Zhou, L.; Montini, T.; Africh, C.; Fornasiero, P.; Comelli, G.; Rosei, R. Electron
localization determines defect formation on ceria substrates. Science 2005, 309, 752 725.
(108) Watkins, M.; Foster, A.; Shluger, A. Hydrogen cycle on CeO2 (111) surfaces: Density functional theory calculations. J. Phys. Chem. C 2007, 2, 15337 15341.
(109) Fronzi, M.; Piccinin, S.; Delley, B.; Traversa, E.; Stampfl, C. Water adsorption on the stoichiometric and reduced CeO2(111) surface: a first-principles investigation. Phys. Chem. Chem. Phys. 2009, 11, 91889199.
(110) Vayssilov, G. N.; Lykhach, Y.; Migani, A.; Staudt, T.; Petrova, G. P.; Tsud, N.; Skála, T.; Bruix, A.; Illas, F.; Prince, K. C.; Matolín, V.; Neyman, K. M.; Libuda, J. Support nanostructure boosts oxygen transfer to catalytically active platinum nanoparticles. Nat. Mater. 2011, 10, 310 315.
(111) Boujana, S.; Dexpert, H.; Majerus, J.; Villain, F.; J, F. L. N. Redox Processes on Pure Ceria and on Rh/CeO2 Catalyst Monitored by X-ray Absorption (Fast Acquisition Mode). J. Phys. Chem. 1994, 21, 5522 5533.
(112) ia and Rh/CeO2 Catalyst Monitored by X- J. Phys. Chem. 1995, 30, 11794 11796.
(113) A. B.; Martínez-Arias, A.; Conesa, J. C. Effects of thermal pretreatment on the redox behaviour of Ce0.5Zr0.5O2: isotopic and spectroscopic
studies. Phys. Chem. Chem. Phys. 2002, 4, 149 159.
(114) 2 ZrO2 mixed oxides. Appl. Catal. B Environ. 1999, 22, L11 L14.
(115) Chen, C.; Yeh, Y.-H.; Cargnello, M.; Murray, C. B.; Fornasiero, P.; Gorte, R. J. Methane Oxidation on Pd@ZrO2/Si Al2O3 Is Enhanced by Surface Reduction of ZrO2. ACS Catal. 2014, 4, 3902 3909.
210
(116) Martín, L. Simulation three-way catalyst ageing. Analysis of two conventional catalyst. Appl. Catal. B Environ. 2003, 44, 41 52.
(117) Heo, I.; Choung, J. W.; Kim, P. S.; Nam, I.-S.; Song, Y. I.; In, C. B.; Yeo, G. K. The alteration of the performance of field-aged Pd-based TWCs towards CO and C3H6 oxidation. Appl. Catal. B Environ. 2009, 92, 114 125.
(118) Larese, C. Deactivation of real three way catalysts by CePO4 formation. Appl. Catal. B Environ. 2003, 40, 305 317.
(119) erium phosphate in aged
automotive catalysts using Raman spectroscopy. Appl. Catal. B Environ. 2003, 41, 269 278.
(120) Benissad- t-Amar, H.; Schouler, M.-C.; Gadelle, P. The role of phosphorus in the growth of vapour-grown carbon fibres obtained by catalytic decomposition of hydrocarbons. Carbon N. Y. 2004, 42, 2163 2168.
(121) Larese, C.; Galisteo, F.; Granados, M.; Mariscal, R.; Fierro, J.; Lambrou, P.; Efstathiou, A. Effects of the CePO4 on the oxygen storage and release properties of CeO2 and Ce0.8Zr0.2O2 solid solution. J. Catal.
2004, 226, 443 456.
(122) Christou, S. Y.; García-Rodríguez, S.; Fierro, J. L. G.; Efstathiou, A. M. Deactivation of Pd/Ce0.5Zr0.5O2 model three-way catalyst by P, Ca and Zn deposition. Appl. Catal. B Environ. 2012, 111 112, 233 245.
(123) Larese, C.; Cabello Galisteo, F.; López Granados, M.; Mariscal López, R.; Fierro, J. L G..; Lambrou, P. S.; Efstathiou, A. M. Effects of calcination temperature on the stability of CePO4 detected in vehicle-aged commercial three-way catalysts. Appl. Catal. B Environ. 2004, 48, 113 123.
(124) Christou, S.; Efstathiou, A. Regeneration of Aged Commercial Three-Way Catalysts Using CI-Containing Reagents. Top. Catal. 2013, 56, 255 260.
(125) Özmen-Monkul, B.; Lerner, M. M. The first graphite intercalation compounds containing tris(pentafluoroethyl)trifluorophosphate. Carbon N. Y. 2010, 48, 3205 3210.
(126) Zhang, Y.; Wang, J.; Zhang, T. Novel Ca-doped CePO4 supported ruthenium catalyst with superior catalytic performance for aerobic oxidation of alcohols. Chem. Commun. 2011, 47, 5307 5309.
(127) Mullins, D. R.; Albrecht, P. M.; Chen, T.-L.; Calaza, F. C.; Biegalski, M. D.; Christen, H. M.; Overbury, S. H. Water Dissociation on CeO2(100) and CeO2(111) Thin Films. J. Phys. Chem. C 2012, 116, 19419 19428.
(128) n, V. CO adsorption on palladium model catalysts: XPS Pd Al2O3 interaction study. Surf. Sci. 2000, 467, 169 176.
(129) He, H.; Gao, C. A General Strategy for the Preparation of Carbon Nanotubes and Graphene Oxide Decorated with PdO Nanoparticles in Water. Molecules 2010, 15, 4679 4694.
(130) Zhang, P.; Sham, T. K. X-Ray Studies of the Structure and Electronic Behavior of Alkanethiolate-Capped Gold Nanoparticles: The Interplay of Size and Surface Effects. Phys. Rev. Lett. 2003, 90, 245502.
(131) Morales, J.; Espinos, J. P.; Caballero, A.; Gonzalez-Elipe, A. R.; Mejias, J. A. XPS Study of Interface and Ligand Effects in Supported Cu2O and CuO Nanometric Particles. J. Phys. Chem. B 2005, 109, 77587765.
(132) Johánek, V.; Stará, I.; Tsud, N.; Veltruská, K.; Matolín, V. CO adsorption on Al2O3-supported Pd clusters: XPS study. Appl. Surf. Sci. 2000, 162, 679 684.
(133) Wang, H.-F.; Kaden, W. E.; Dowler, R.; Sterrer, M.; Freund, H.-J. Model oxide-supported metal catalysts - comparison of ultrahigh vacuum and solution based preparation of Pd nanoparticles on a single-crystalline oxide substrate. Phys. Chem. Chem. Phys. 2012, 14, 11525 11533.
(134) Penner, S.; Bera, P.; Pedersen, S.; Ngo, L. T.; Harris, J. J. W.; Campbell, C. T. Interactions of O2 with Pd
211
-Al2O3(0001) at Low and High O2 Pressures. J. Phys. Chem. B 2006, 110, 2457724584.
(135) Lin, R.; Freemantle, R. G.; Kelly, N. M.; Fielitz, T. R.; Obare, S. O.; Ofoli, R. Y. In situ immobilization of palladium nanoparticles in microfluidic reactors and assessment of their catalytic activity. Nanotechnology 2010, 21, 325605.
(136) Gélin, P.; Urfels, L.; Primet, M.; Tena, E. Complete oxidation of methane at low temperature over Pt and Pd catalysts for the abatement of lean-burn natural gas fuelled vehicles emissions: influence of water and sulphur containing compounds. Catal. Today 2003, 83, 45 57.
(137) United States Environmental Protection Agency. EPA420-F-06-064; 2006.
(138) Colussi, S.; Arosio, F.; Montanari, T.; Busca, G.; Groppi, G.; Trovarelli, A. Study of sulfur poisoning on Pd/Al2O3 and Pd/CeO2/Al2O3 methane combustion catalysts. Catal. Today 2010, 155, 59 65.
(139) Luo, T.; Vohs, J. M.; Gorte, R. J. An Examination of Sulfur Poisoning on Pd/Ceria Catalysts. J. Alloys Compd. 2002, 210, 397 404.
(140) Rodriguez, J. A.; Jirsak, T.; Freitag, A.; Hanson, J. C.; Larese, J. Z.; Chaturvedi, S. Interaction of SO2 with CeO2 and Cu/CeO2 catalysts: photoemission, XANES and TPD studies. Catal. Letters 1999, 62, 113 119.
(141) Bazin, P.; Saur, O.; Lavalley, J. C.; Blanchard, G.; Visciglio, V.; Touret, O. Influence of platinum on ceria sulfation. Appl. Catal. B Environ. 1997, 13, 265 274.
(142) Luo, T.; Gorte, R. J. Characterization of SO2-poisoned ceria-zirconia mixed oxides. Appl. Catal. B Environ. 2004, 53, 77 85.
(143) Romano, E. J.; Schulz, K. H. A XPS investigation of SO2 adsorption on ceria zirconia mixed-metal oxides. Appl. Surf. Sci. 2005, 246, 262 270.
(144) 2 on CO and C3H6 oxidation over CeO2 and Ce0.75Zr0.25O2. Appl. Catal. B Environ. 2003, 45, 135 145.
(145) Waqif, M.; Bazin, P.; Saur, O.; Lavalley, J. C.; Blanchard, G.; Touret, O. Study of ceria sulfation. Appl. Catal.
B Environ. 1997, 11, 193 205.
(146) Nelson, A. E.; Schulz, K. H. Surface chemistry and microstructural analysis of CexZr O model catalyst surfaces. Appl. Surf. Sci. 2003, 210, 206 221.
(147) aluminas. Acta Crystallogr. Sect. B Struct. Sci. 1991, 47, 617 630.
(148) Colón, G.; Valdivieso, F.; Pijolat, M.; Baker, R. T.; Calvino, J. J.; Bernal, S. Textural and phase stability of CexZr O2 mixed oxides under high temperature oxidising conditions. Catal. Today 1999, 50, 271284.
(149) Montini, T.; Melchionna, M.; Monai, M.; Fornasiero, P. Fundamentals and Catalytic Applications of CeO2-Based Materials. Chem. Rev. 2016.
(150) Zhang, S.; Chen, C.; Cargnello, M.; Fornasiero, P.; Gorte, R. J.; Graham, G. W.; Pan, X. Dynamic structural evolution of supported palladium-ceria core-shell catalysts revealed by in situ electron microscopy. Nat. Commun. 2015, 6, 7778.
(151) Burch, R.; Crittle, D. J.; Southward, B. W. L.; Sullivan, J. A. The effect of SO2 on the activity of Pd-based catalysts in methane combustion. Catal. Letters 2001, 72, 153 155.
(152) J.; Fornasiero, P. Phosphorus poisoning during wet oxidation of methane over Pd@CeO2/graphite model catalysts. Appl. Catal. B Environ. 2015.
(153) Priolkar, K. R.; Bera, P.; Sarode, P. R.; Hegde, M. S.; Emura, S.; Kumashiro, R.; Lalla, N. P. Formation of Ce1-
212
xPdxO2- Solid Solution in Combustion-Synthesized Pd/CeO2 Catalyst: XRD, XPS, and EXAFS Investigation. Chem. Mater. 2002, 14, 2120 2128.
(154) Hiley, C. I.; Fisher, J. M.; Thompsett, D.; Kashtiban, R. J.; Sloan, J.; Walton, R. I. Incorporation of square-planar Pd2+ in fluorite CeO2: hydrothermal preparation, local structure, redox properties and stability. J. Mater. Chem. A 2015, 3, 13072 13079.
(155) Venezia, A.; Murania, R.; Pantaleo, G.; Deganello, G. Pd and PdAu on mesoporous silica for methane oxidation: Effect of SO2. J. Catal. 2007, 251, 94 102.
(156) Liotta, L. F.; Di Carlo, G.; Pantaleo, G.; Venezia, A. M.; Deganello, G. Insights into SO2 Interaction with Pd/Co3O4 CeO2 Catalysts for Methane Oxidation. Top. Catal. 2009, 52, 1989 1994.
(157) Sharma, H. N.; Sharma, V.; Mhadeshwar, A. B.; Ramprasad, R. Why Pt Survives but Pd Suffers From SOx Poisoning? J. Phys. Chem. Lett. 2015, 6, 1140 1148.
(158) Yeh, J. J.; Lindau, I. Atomic subshell photoionization cross sections and asymmetry parameters: 1 < Z <103. At. Data Nucl. Data Tables 1985, 32, 1 155.
(159) Ferrizz, R. M.; Gorte, R. J.; Vohs, J. M. TPD and XPS Investigation of the Interaction of SO2 with Model Ceria Catalysts. Catal. Letters 2002, 82, 123 129.
(160) Dumesic, J. A.; Huber, G. W. Catalyst Design with Atomic Layer Deposition. ACS Catal. 2015, 5, 18041825.
(161) Burch, R.; Urbano, F. J. Investigation of the active state of supported palladium catalysts in the combustion of methane. Appl. Catal. A Gen. 1995, 124, 121 138.
(162) Cargnello, M.; Wieder, N. L.; Canton, P.; Montini, T.; Giambastiani, G.; Benedetti, A.; Gorte, R. J.; Fornasiero, P. A Versatile Approach to the Synthesis of Functionalized Thiol-Protected Palladium Nanoparticles. Chem. Mater. 2011, 23, 3961 3969.
(163) Nanda, S.; Mohammad, J.; Reddy, S. N.; Kozinski, J. A.; Dalai, A. K. Pathways of lignocellulosic biomass conversion to renewable fuels. Biomass Convers. Biorefinery 2014, 4, 157 191.
(164) Fukuoka, A.; Dhepe, P. L. Catalytic conversion of cellulose into sugar alcohols. Angew. Chem. Int. Ed. Engl. 2006, 45, 5161 5163.
(165) Hegner, J.; Pereira, K. C.; DeBoef, B.; Lucht, B. L. Conversion of cellulose to glucose and levulinic acid via solid-supported acid catalysis; 2010; Vol. 51.
(166) Gürbüz, E. I.; Wettstein, S. G.; Dumesic, J. A. Conversion of hemicellulose to furfural and levulinic acid using biphasic reactors with alkylphenol solvents. ChemSusChem 2012, 5, 383 387.
(167) Lanzafame, P.; Temi, D. M.; Perathoner, S.; Spadaro, A. N.; Centi, G. Direct conversion of cellulose to glucose and valuable intermediates in mild reaction conditions over solid acid catalysts. In Catalysis Today; 2012; Vol. 179, pp. 178 184.
(168) Alonso, D. M.; Wettstein, S. G.; Mellmer, M. A.; Gurbuz, E. I.; Dumesic, J. A. Integrated conversion of hemicellulose and cellulose from lignocellulosic biomass. Energy Environ. Sci. 2013, 6, 76 80.
(169) Sutton, A. D.; Waldie, F. D.; Wu, R.; Schlaf, M.; of bioderived furans into alkanes. Nat. Chem. 2013, 5, 428 432.
(170) Asadullah, M.; Rahman, M. A.; Ali, M. M.; Rahman, M. S.; Motin, M. A.; Sultan, M. B.; Alam, M. R. Production of bio-oil from fixed bed pyrolysis of bagasse. Fuel 2007, 86, 2514 2520.
(171) Demirbas, A. The influence of temperature on the yields of compounds existing in bio-oils obtained from biomass samples via pyrolysis. Fuel Process. Technol. 2007, 88, 591 597.
(172) Elliott, D. C.; Hart, T. R. Catalytic Hydroprocessing of Chemical Models for Bio-oil. Energy & Fuels 2009,
213
23, 631 637.
(173) Serrano-Ruiz, J. C.; West, R. M.; Dumesic, J. A. Catalytic Conversion of Renewable Biomass Resources to
Fuels and Chemicals. Annu. Rev. Chem. Biomol. Eng. 2010, 1, 79 100.
(174) Nakagawa, Y.; Tamura, M.; Tomishige, K. Catalytic Reduction of Biomass-Derived Furanic Compounds with Hydrogen. ACS Catal. 2013, 3, 2655 2668.
(175) Román-Leshkov, Y.; Barrett, C. J.; Liu, Z. Y.; Dumesic, J. A. Production of dimethylfuran for liquid fuels from biomass-derived carbohydrates. Nature 2007, 447, 982 985.
(176) Wang, T.; Nolte, M. W.; Shanks, B. H. Catalytic dehydration of C6 carbohydrates for the production of hydroxymethylfurfural (HMF) as a versatile platform chemical. Green Chem. 2014, 16, 548 572.
(177) Dashtban, M.; Gilbert, A.; Fatehi, P. Recent advancements in the production of hydroxymethylfurfural.
RSC Adv. 2014, 4, 2037 2050.
(178) Torres, A. I.; Daoutidis, P.; Tsapatsis, M. Continuous production of 5-hydroxymethylfurfural from fructose: a design case study. Energy Environ. Sci. 2010, 3, 1560.
(179) Chen, G.; Shen, Y.; Zhang, Q.; Yao, M.; Zheng, Z.; Liu, H. Experimental study on combustion and emission characteristics of a diesel engine fueled with 2,5-dimethylfuran diesel, n-butanol diesel and gasoline diesel blends. Energy 2013, 54, 333 342.
(180) Green, S. K.; Patet, R. E.; Nikbin, N.; Williams, C. L.; Chang, C.-C.; Yu, J.; Gorte, R. J.; Caratzoulas, S.; Fan, W.; Vlachos, D. G.; Dauenhauer, P. J. Diels Alder cycloaddition of 2-methylfuran and ethylene for renewable toluene. Appl. Catal. B Environ. 2016, 180, 487 496.
(181) Williams, C. L.; Chang, C.-C.; Do, P.; Nikbin, N.; Caratzoulas, S.; Vlachos, D. G.; Lobo, R. F.; Fan, W.; Dauenhauer, P. J. Cycloaddition of Biomass-Derived Furans for Catalytic Production of Renewable p -Xylene. ACS Catal. 2012, 2, 935 939.
(182) Sitthisa, S.; Resasco, D. E. Hydrodeoxygenation of Furfural Over Supported Metal Catalysts: A Comparative Study of Cu, Pd and Ni. Catal. Letters 2011, 141, 784 791.
(183) Pushkarev, V. V; Musselwhite, N.; An, K.; Alayoglu, S.; Somorjai, G. A. High Structure Sensitivity of Vapor-Phase Furfural Decarbonylation/Hydrogenation Reaction Network as a Function of Size and Shape of Pt Nanoparticles. Nano Lett. 2012, 12, 5196 5201.
(184) Pang, S. H.; Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Directing reaction pathways by catalyst
active-site selection using self-assembled monolayers. Nat. Commun. 2013, 4.
(185) Luo, J.; Arroyo-Ramírez, L.; Wei, J.; Yun, H.; Murray, C. B.; Gorte, R. J. Comparison of HMF hydrodeoxygenation over different metal catalysts in a continuous flow reactor. Appl. Catal. A Gen.
2015, 508, 86 93.
(186) Hu, L.; Tang, X.; Xu, J.; Wu, Z.; Lin, L.; Liu, S. Selective transformation of 5-hydroxymethylfurfural into the liquid fuel 2,5-dimethylfuran over carbon-supported ruthenium. Ind. Eng. Chem. Res. 2014, 53, 3056 3064.
(187) Saha, B.; Bohn, C. M.; Abu-Omar, M. M. Zinc-assisted hydrodeoxygenation of biomass-derived 5-hydroxymethylfurfural to 2,5-Dimethylfuran. ChemSusChem 2014, 7, 3095 3101.
(188) Wang, G.-H.; Hilgert, J.; Richter, F. H.; Wang, F.; Bongard, H.-J.; Spliethoff, B.; Weidenthaler, C.; Schüth, F. Platinum-cobalt bimetallic nanoparticles in hollow carbon nanospheres for hydrogenolysis of 5-hydroxymethylfurfural. Nat. Mater. 2014, 13, 293 300.
(189) Wang, S.; Vorotnikov, V.; Vlachos, D. G. Coverage-Induced Conformational Effects on Activity and Selectivity: Hydrogenation and Decarbonylation of Furfural on Pd(111). ACS Catal. 2015, 5, 104 112.
(190) Bradley, M. K.; Robinson, J.; Woodruff, D. P. The structure and bonding of furan on Pd(111). Surf. Sci.
214
2010, 604, 920 925.
(191) Yan, K.; Chen, A. Selective hydrogenation of furfural and levulinic acid to biofuels on the ecofriendly
Cu Fe catalyst. Fuel 2014, 115, 101 108.
(192) Hronec, M.; Fulajtarová, K. Selective transformation of furfural to cyclopentanone. Catal. Commun. 2012, 24, 100 104.
(193) Nishimura, S.; Ikeda, N.; Ebitani, K. Selective hydrogenation of biomass-derived 5-hydroxymethylfurfural (HMF) to 2,5-dimethylfuran (DMF) under atmospheric hydrogen pressure over carbon supported PdAu bimetallic catalyst. In Catalysis Today; 2014; Vol. 232, pp. 89 98.
(194) Myint, M.; Yan, Y.; Chen, J. G. Reaction pathways of propanal and 1-propanol on Fe/Ni(111) and Cu/Ni(111) bimetallic surfaces. J. Phys. Chem. C 2014, 118, 11340 11349.
(195) Lee, J.; Kim, Y. T.; Huber, G. W. Aqueous-phase hydrogenation and hydrodeoxygenation of biomass-derived oxygenates with bimetallic catalysts. Green Chem. 2014, 16, 708.
(196) Chidambaram, M.; Bell, A. T. A two-step approach for the catalytic conversion of glucose to 2,5-
dimethylfuran in ionic liquids. Green Chem. 2010, 12, 1253.
(197) Thananatthanachon, T.; Rauchfuss, T. B. Efficient Production of the Liquid Fuel 2,5-Dimethylfuran from Fructose Using Formic Acid as a Reagent. Angew. Chemie Int. Ed. 2010, 49, 6616 6618.
(198) Jae, J.; Zheng, W.; Lobo, R. F.; Vlachos, D. G. Production of Dimethylfuran from Hydroxymethylfurfural through Catalytic Transfer Hydrogenation with Ruthenium Supported on Carbon. ChemSusChem 2013, 6, 1158 1162.
(199) Shevchenko, E. V; Talapin, D. V; Rogach, A. L.; Kornowski, A.; Haase, M.; Weller, H. Colloidal synthesis and self-assembly of CoPt3 nanocrystals. J. Am. Chem. Soc. 2002, 124, 11480 11485.
(200) Gordon, T. R.; Cargnello, M.; Paik, T.; Mangolini, F.; Weber, R. T.; Fornasiero, P.; Murray, C. B. Nonaqueous Synthesis of TiO2 Nanocrystals Using TiF4 to Engineer Morphology, Oxygen Vacancy Concentration, and Photocatalytic Activity. J. Am. Chem. Soc. 2012, 134, 6751 6761.
(201) Fievet, F.; Lagier, J.; Blin, B.; Beaudoin, B.; Figlarz, M. Homogeneous and heterogeneous nucleations in the polyol process for the preparation of micron and submicron size metal particles. Solid State Ionics 1989, 32 33, 198 205.
(202) Yu, Y.; Yang, W.; Sun, X.; Zhu, W.; Li, X. Z.; Sellmyer, D. J.; Sun, S. Monodisperse MPt (M = Fe, Co, Ni, Cu, Zn) nanoparticles prepared from a facile oleylamine reduction of metal salts. Nano Lett. 2014, 14, 2778 2782.
(203) Wang, D.; Xin, H. L.; Hovden, R.; Wang, H.; Yu, Y.; Muller, D. A.; Di Salvo, F. J.; Abruña, H. D. Structurally ordered intermetallic platinum cobalt core shell nanoparticles with enhanced activity and stability as oxygen reduction electrocatalysts. Nat. Mater. 2012, 12, 81 87.
(204) Ng, Y. H.; Ikeda, S.; Harada, T.; Sakata, T.; Mori, H.; Takaoka, A.; Matsumura, M. High sintering resistance of platinum nanoparticles embedded in a microporous hollow carbon shell fabricated through a photocatalytic reaction. Langmuir 2008, 24, 6307 6312.
(205) Lonergan, W. W.; Vlachos, D. G.; Chen, J. G. Correlating extent of Pt Ni bond formation with low-temperature hydrogenation of benzene and 1,3-butadiene over supported Pt/Ni bimetallic catalysts.
J. Catal. 2010, 271, 239 250.
(206) Menning, C. A.; Chen, J. G. Thermodynamics and kinetics of oxygen-induced segregation of 3d metals in Pt-3d-Pt (111) and Pt-3d-Pt (100) bimetallic structures. J. Chem. Phys. 2008, 128, 164703 164703.
(207) De Santis, M.; Buchsbaum, A.; Varga, P.; Schmid, M. Growth of ultrathin cobalt oxide films on Pt (111). Phys. Rev. B 2011, 84, 125430.
215
(208) Blaylock, D. W.; Zhu, Y.-A.; Green, W. H. Computational investigation of the thermochemistry and kinetics of steam methane reforming over a multi-faceted nickel catalyst. Top. Catal. 2011, 54, 828844.
(209) Zhu, Z.; Tao, F.; Zheng, F.; Chang, R.; Li, Y.; Heinke, L.; Liu, Z.; Salmeron, M.; Somorjai, G. A. Formation of nanometer-sized surface platinum oxide clusters on a stepped Pt (557) single crystal surface induced by oxygen: a high-pressure STM and ambient-pressure XPS study. Nano Lett. 2012, 12, 1491 1497.
(210) Luo, J.; Yun, H.; Mironenko, A. V.; Goulas, K.; Lee, J. D.; Monai, M.; Wang, C.; Vorotnikov, V.; Murray, C. B.; Vlachos, D. G.; Fornasiero, P.; Gorte, R. J. Mechanisms for High Selectivity in the Hydrodeoxygenation of 5-Hydroxymethylfurfural over PtCo Nanocrystals. ACS Catal. 2016, 6, 4095 4104.
(211) Knudsen, J.; Merte, L. R.; Grabow, L. C.; Eichhorn, F. M.; Porsgaard, S.; Zeuthen, H.; Vang, R. T.; Lægsgaard, E.; Mavrikakis, M.; Besenbacher, F. Reduction of FeO/Pt (111) thin films by exposure to atomic hydrogen. Surf. Sci. 2010, 604, 11 20.
(212) Tengco, J. M. M.; Lugo-José, Y. K.; Monnier, J. R.; Regalbuto, J. R. Chemisorption XRD particle size
discrepancy of carbon supported palladium: Carbon decoration of Pd? Catal. Today 2015, 246, 9 14.
(213) Scholten, J. J. F.; Pijpers, A. P.; Hustings, A. M. L. Surface Characterization of Supported and Nonsupported Hydrogenation Catalysts. Catal. Rev. Sci. Eng. 1985, 27, 151 206.
(214) Tuokko, S.; Pihko, P. M.; Honkala, K. First Principles Calculations for Hydrogenation of Acrolein on Pd and Pt: Chemoselectivity Depends on Steric Effects on the Surface. Angew. Chemie Int. Ed. 2016.
(215) Loffreda, D.; Delbecq, F.; Vigné, F.; Sautet, P. Chemo-regioselectivity in heterogeneous catalysis: Competitive routes for CO and CC hydrogenations from a theoretical approach. J. Am. Chem. Soc. 2006, 128, 1316 1323.
(216) Caratzoulas, S.; Vlachos, D. G. Converting fructose to 5-hydroxymethylfurfural: a quantum mechanics/molecular mechanics study of the mechanism and energetics. Carbohydr. Res. 2011, 346, 664 672.
(217) Mironenko, A. V.; Gilkey, M. J.; Panagiotopoulou, P.; Facas, G.; Vlachos, D. G.; Xu, B. Ring Activation of Furanic Compounds on Ruthenium-Based Catalysts. J. Phys. Chem. C 2015, 119, 6075 6085.
(218) Wang, S.; Vorotnikov, V.; Vlachos, D. G. Coverage-Induced Conformational Effects on Activity and
Selectivity: Hydrogenation and Decarbonylation of Furfural on Pd (111). Acs Catal. 2014, 5, 104 112.
(219) Vorotnikov, V.; Mpourmpakis, G.; Vlachos, D. G. DFT study of furfural conversion to furan, furfuryl alcohol, and 2-methylfuran on Pd (111). Acs Catal. 2012, 2, 2496 2504.
(220) Wang, S.; Vorotnikov, V.; Vlachos, D. G. A DFT study of furan hydrogenation and ring opening on Pd (111). Green Chem. 2014, 16, 736 747.
(221) Pang, S. H.; Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Effects of Thiol Modifiers on the Kinetics of Furfural Hydrogenation over Pd Catalysts. Acs Catal. 2014, 4, 3123 3131.
(222) Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Controlling the surface environment of
heterogeneous catalysts using self-assembled monolayers. Acc. Chem. Res. 2014, 47, 1438 1445.
(223) McManus, J. R.; Martono, E.; Vohs, J. M. Selective Deoxygenation of Aldehydes: The Reaction of Acetaldehyde and Glycolaldehyde on Zn/Pt(111) Bimetallic Surfaces. ACS Catal. 2013, 3, 1739 1750.
(224) Shi, D.; Vohs, J. M. Deoxygenation of Biomass-Derived Oxygenates: Reaction of Furfural on Zn-Modified Pt(111). ACS Catal. 2015, 5, 2177 2183.
(225) Yu, W.; Xiong, K.; Ji, N.; Porosoff, M. D.; Chen, J. G. Theoretical and experimental studies of the adsorption geometry and reaction pathways of furfural over FeNi bimetallic model surfaces and supported catalysts. J. Catal. 2014, 317, 253 262
216
(226) Sitthisa, S.; Pham, T.; Prasomsri, T.; Sooknoi, T.; Mallinson, R. G.; Resasco, D. E. Conversion of furfural and 2-methylpentanal on Pd/SiO2 and Pd Cu/SiO2 catalysts. J. Catal. 2011, 280, 17 27.
(227) Sitthisa, S.; Sooknoi, T.; Ma, Y.; Balbuena, P. B.; Resasco, D. E. Kinetics and mechanism of hydrogenation of furfural on Cu/SiO2 catalysts. J. Catal. 2011, 277, 1 13.
(228) Ahrenstorf, K.; Albrecht, O.; Heller, H.; Kornowski, A.; Görlitz, D.; Weller, H. Colloidal Synthesis of NixPt Nanoparticles with Tuneable Composition and Size. Small 2007, 3, 271 274.
(229) Murray, C. B.; Sun, S.; Doyle, H.; Betley, T. Monodisperse 3d transition-metal (Co, Ni, Fe) nanoparticles and their assembly intonanoparticle superlattices. Mrs Bull. 2001, 26, 985 991.
(230) Phys. Rev. A 1991, 43, 3161.
(231) Kang, Y.; Ye, X.; Murray, C. B. Size- and Shape-Selective Synthesis of Metal Nanocrystals and Nanowires
Using CO as a Reducing Agent. Angew. Chemie Int. Ed. 2010, 49, 6156 6159.
(232) Luo, J.; Lee, J. D.; Yun, H.; Wang, C.; Monai, M.; Murray, C. B.; Fornasiero, P.; Gorte, R. J. Base metal-Pt alloys: A general route to high selectivity and stability in the production of biofuels from HMF. Appl.
Catal. B Environ. 2016, 199, 439 446.
(233) Holade, Y.; Morais, C.; Servat, K.; Napporn, T. W.; Kokoh, K. B. Enhancing the available specific surface area of carbon supports to boost the electroactivity of nanostructured Pt catalysts. Phys. Chem. Chem.
Phys. 2014, 16, 25609 25620.
(234) Yu, L.; He, L.; Chen, J.; Zheng, J.; Ye, L.; Lin, H.; Yuan, Y. Robust and Recyclable Nonprecious Bimetallic Nanoparticles on Carbon Nanotubes for the Hydrogenation and Hydrogenolysis of 5-Hydroxymethylfurfural. ChemCatChem 2015, 7, 1701 1707.
(235) Huang, Y.-B.; Chen, M.-Y.; Yan, L.; Guo, Q.-X.; Fu, Y. Nickel-Tungsten Carbide Catalysts for the Production of 2,5-Dimethylfuran from Biomass-Derived Molecules. ChemSusChem 2014, 7, 10681072.
(236) Chen, B.; Li, F.; Huang, Z.; Yuan, G. Carbon-coated Cu-Co bimetallic nanoparticles as selective and recyclable catalysts for production of biofuel 2,5-dimethylfuran. Appl. Catal. B Environ. 2017, 200, 192199.
(237) Huba, Z. J.; Carpenter, E. E. Ethanol assisted reduction and nucleation of ferromagnetic Co and Ni nanocrystalline particles. CrystEngComm 2013, 15, 8919.
(238) Pellicer, E.; Varea, A.; Sivaraman, K. M.; Pané, S.; Suriñach, S.; Baró, M. D.; Nogués, J.; Nelson, B. J.; Sort, J. Grain Boundary Segregation and Interdiffusion Effects in Nickel Copper Alloys: An Effective Means to Improve the Thermal Stability of Nanocrystalline Nickel. ACS Appl. Mater. Interfaces 2011, 3, 22652274.
(239) Luo, J.; Yu, J.; Gorte, R. J.; Mahmoud, E.; Vlachos, D. G.; Smith, M. A. The effect of oxide acidity on HMF etherification. Catal. Sci. Technol. 2014, 4, 3074 3081.
(240) Yao, Y.; Goodman, D. W. Direct evidence of hydrogen spillover from Ni to Cu on Ni Cu bimetallic
catalysts. J. Mol. Catal. A Chem. 2014, 383 384, 239 242.
(241) Sinfelt, J. H. Catalysis by alloys and bimetallic clusters. Acc. Chem. Res. 1977, 10, 15 20.
(242) Xiong, K.; Wan, W.; Chen, J. G. Reaction pathways of furfural, furfuryl alcohol and 2-methylfuran on Cu(111) and NiCu bimetallic surfaces. Surf. Sci. 2016, 652, 91 97.
(243) Bakhmutsky, K.; Wieder, N. L.; Baldassare, T.; Smith, M. A.; Gorte, R. J. A thermodynamic study of the redox properties of supported Co particles. Appl. Catal. A Gen. 2011, 397, 266 271.
(244) Shi, D.; Arroyo-Ramírez, L.; Vohs, J. M. The use of bimetallics to control the selectivity for the upgrading of lignin-derived oxygenates: Reaction of anisole on Pt and PtZn catalysts. J. Catal. 2016,
217
340, 219 226.
(245) Press, R. J.; Santhanam, K. S. V.; Miri, M. J.; Bailey, A. V.; Takacs, G. A. Introduction to Hydrogen Technology; Wiley, 2008.
(246) Armaroli, N.; Balzani, V. Solar Electricity and Solar Fuels: Status and Perspectives in the Context of the Energy Transition. Chem. Eur. J. 2016, 22, 32 57.
(247) Kudo, A.; Miseki, Y. Heterogeneous photocatalyst materials for water splitting. Chem. Soc. Rev. 2009, 38, 253 278.
(248) Zaleska, A. Doped-TiO2: A Review. Recent Patents Eng. 2008, 2, 157 164.
(249) Tanaka, A.; Sakaguchi, S.; Hashimoto, K.; Kominami, H. Preparation of Au/TiO2 with Metal Cocatalysts Exhibiting Strong Surface Plasmon Resonance Effective for Photoinduced Hydrogen Formation under Irradiation of Visible Light. ACS Catal. 2013, 3, 79 85.
(250) Chen, X.; Liu, L.; Yu, P. Y.; Mao, S. S. Increasing Solar Absorption for Photocatalysis with Black Hydrogenated Titanium Dioxide Nanocrystals. Science 2011, 331, 746 750.
(251) Zhang, X.; Peng, T.; Song, S. Recent advances in dye-sensitized semiconductor systems for photocatalytic hydrogen production. J. Mater. Chem. A 2016, 4, 2365 2402.
(252) Chen, X.; Shen, S.; Guo, L.; Mao, S. S. Semiconductor-based Photocatalytic Hydrogen Generation. Chem. Rev. 2010, 110, 6503 6570.
(253) Qureshi, M.; Takanabe, K. Insights on Measuring and Reporting Heterogeneous Photocatalysis: Efficiency Definitions and Setup Examples. Chem. Mater. 2017, 29, 158 167.
(254) Youngblood, W. J.; Lee, S.-H. A.; Kobayashi, Y.; Hernandez-Pagan, E. A.; Hoertz, P. G.; Moore, T. A.; Moore, A. L.; Gust, D.; Mallouk, T. E. Photoassisted Overall Water Splitting in a Visible Light-Absorbing
Dye-Sensitized Photoelectrochemical Cell. J. Am. Chem. Soc. 2009, 131, 926 927.
(255) Abbotto, A.; Manfredi, N.; Marinzi, C.; De Angelis, F.; Mosconi, E.; Yum, J.-H.; Xianxi, Z.; Nazeeruddin, M. K.; Gratzel, M. Di-branched di-anchoring organic dyes for dye-sensitized solar cells. Energy Environ. Sci.
2009, 2, 1094 1101.
(256) Lin, R. Y.-Y.; Wu, F.-L.; Li, C.-T.; Chen, P.-Y.; Ho, K.-C.; Lin, J. T. High-Performance Aqueous/Organic Dye-Sensitized Solar Cells Based on Sensitizers Containing Triethylene Oxide Methyl Ether. ChemSusChem
2015, 8, 2503 2513.
(257) Lee, S.-H.; Park, Y.; Wee, K.-R.; Son, H.-J.; Cho, D. W.; Pac, C.; Choi, W.; Kang, S. O. Significance of Hydrophilic Characters of Organic Dyes in Visible-Light Hydrogen Generation Based on TiO2. Org. Lett.
2010, 12, 460 463.
(258) Pescitelli, G.; Omar, O. H.; Operamolla, A.; Farinola, G. M.; Di Bari, L. Chiroptical Properties of Glucose-Substituted Poly(p-phenylene-ethynylene)s in Solution and Aggregate State. Macromolecules 2012, 45, 9626 9630.
(259) Haldon, E.; Nicasio, M. C.; Perez, P. J. Copper-catalysed azide-alkyne cycloadditions (CuAAC): an update. Org. Biomol. Chem. 2015, 13, 9528 9550.
(260) Chabre, Y. M.; Roy, R. Multivalent glycoconjugate syntheses and applications using aromatic scaffolds. Chem. Soc. Rev. 2013, 42, 4657 4708.
(261) Tornøe, C. W.; Christensen, C.; Meldal, M. Peptidotriazoles on Solid Phase: [1,2,3]-Triazoles by Regiospecific Copper(I)-Catalyzed 1,3-Dipolar Cycloadditions of Terminal Alkynes to Azides. J. Org. Chem. 2002, 67, 3057 3064.
(262) Manfredi, N.; Cecconi, B.; Calabrese, V.; Minotti, A.; Peri, F.; Ruffo, R.; Monai, M.; Romero-Ocana, I.; Montini, T.; Fornasiero, P.; Abbotto, A. Dye-sensitized photocatalytic hydrogen production: distinct
218
activity in a glucose derivative of a phenothiazine dye. Chem. Commun. 2016.
(263) Romero Ocaña, I.; Beltram, A.; Delgado Jaén, J. J.; Adami, G.; Montini, T.; Fornasiero, P. Photocatalytic H2 production by ethanol photodehydrogenation: Effect of anatase/brookite nanocomposites composition. Inorganica Chim. Acta 2015, 431, 197 205.
(264) Varma, A. J.; Kennedy, J. F.; Galgali, P. Synthetic polymers functionalized by carbohydrates: a review. Carbohydr. Polym. 2004, 56, 429 445.
(265) Li, J.; Grimsdale, A. C. Carbazole-based polymers for organic photovoltaic devices. Chem. Soc. Rev. 2010, 39, 2399 2410.
(266) Schmidt, A. W.; Reddy, K. R.; Knölker, H.-J. Occurrence, Biogenesis, and Synthesis of Biologically Active Carbazole Alkaloids. Chem. Rev. 2012, 112, 3193 3328.
(267) Manfredi, N.; Cecconi, B.; Abbotto, A. Multi-Branched Multi-Anchoring Metal-Free Dyes for Dye-Sensitized Solar Cells. European J. Org. Chem. 2014, 2014, 7069 7086.
(268) Chen, C.; Liao, J.-Y.; Chi, Z.; Xu, B.; Zhang, X.; Kuang, D.-B.; Zhang, Y.; Liu, S.; Xu, J. Metal-free organic dyes derived from triphenylethylene for dye-sensitized solar cells: tuning of the performance by phenothiazine and carbazole. J. Mater. Chem. 2012, 22, 8994 9005.
(269) Gupta, K. S. V; Suresh, T.; Singh, S. P.; Islam, A.; Han, L.; Chandrasekharam, M. Carbazole based A- -D- -A dyes with double electron acceptor for dye-sensitized solar cell. Org. Electron. 2014, 15, 266 275.
(270) Abe, R.; Shinmei, K.; Koumura, N.; Hara, K.; Ohtani, B. Visible-Light-Induced Water Splitting Based on Two-Step Photoexcitation between Dye-Sensitized Layered Niobate and Tungsten Oxide Photocatalysts in the Presence of a Triiodide/Iodide Shuttle Redox Mediator. J. Am. Chem. Soc. 2013, 135, 16872 16884.
(271) Sprick, R. S.; Bonillo, B.; Clowes, R.; Guiglion, P.; Brownbill, N. J.; Slater, B. J.; Blanc, F.; Zwijnenburg, M. A.; Adams, D. J.; Cooper, A. I. Visible-Light-Driven Hydrogen Evolution Using Planarized Conjugated
Polymer Photocatalysts. Angew. Chemie Int. Ed. 2016, 55, 1792 1796.
(272) Su, C.; Tandiana, R.; Tian, B.; Sengupta, A.; Tang, W.; Su, J.; Loh, K. P. Visible-Light Photocatalysis of Aerobic Oxidation Reactions Using Carbazolic Conjugated Microporous Polymers. ACS Catal. 2016, 6, 3594 3599.
(273) McDowell, J. J. H. The crystal and molecular structure of phenothiazine. Acta Crystallogr. Sect. B 1976, 32, 5 10.
(274) Comprehensive Heterocyclic Chemistry: Five-membered rings with two or more nitrogen atoms; Katritzky, A. R., Ed.; Oxford, 1984.
(275) Grisorio, R.; De Marco, L.; Allegretta, G.; Giannuzzi, R.; Suranna, G. P.; Manca, M.; Mastrorilli, P.; Gigli, G. Anchoring stability and photovoltaic properties of new D(- -A)2 dyes for dye-sensitized solar cell applications. Dyes Pigments 2013, 98, 221 231.
(276) Tauc, J. Optical properties and electronic structure of amorphous Ge and Si. Mater. Res. Bull. 1968, 3, 37 46.
(277) Dye-sensitized Solar Cells; Kalyanasundaram, K., Ed.; CRC Press, 2010.
(278) Park, H.; Kim, H.; Moon, G.; Choi, W. Photoinduced charge transfer processes in solar photocatalysis based on modified TiO2. Energy Environ. Sci. 2016, 9, 411 433.
(279) Zhang, G.; Kim, G.; Choi, W. Visible light driven photocatalysis mediated via ligand-to-metal charge transfer (LMCT): an alternative approach to solar activation of titania. Energy Environ. Sci. 2014, 7, 954966.
(280) Hung, W.-I.; Liao, Y.-Y.; Hsu, C.-Y.; Chou, H.-H.; Lee, T.-H.; Kao, W.-S.; Lin, J. T. High-Performance Dye-
219
Sensitized Solar Cells Based on Phenothiazine Dyes Containing Double Anchors and Thiophene Spacers. Chem. Asian J. 2014, 9, 357 366.
(281) Al-Azri, Z. H. N.; Jovic, V.; Chen, W.-T.; Sun-Waterhouse, D.; Metson, J. B.; Waterhouse, G. I. N. Performance evaluation of Pd/TiO2 and Pt/TiO2 photocatalysts for hydrogen production from ethanol-water mixtures. Int. J. Nanotechnol. 2014, 11, 695 703.
(282) Caravaca, A.; Jones, W.; Hardacre, C.; Bowker, M. H2 production by the photocatalytic reforming of
cellulose and raw biomass using Ni, Pd, Pt and Au on titania. Proc. Math. Phys. Eng. Sci. 2016, 472, 20160054.
(283) Montini, T.; Gombac, V.; Sordelli, L.; Delgado, J. J.; Chen, X.; Adami, G.; Fornasiero, P. Nanostructured Cu/TiO2 Photocatalysts for H2 Production from Ethanol and Glycerol Aqueous Solutions. ChemCatChem 2011, 3, 574 577.
(284) Weissermel, K.; Arpe, H.-J. Industrial Organic Chemistry; 4th ed.; WILEY-VCH Verlag, 2003.
(285) Montini, T.; Monai, M.; Beltram, A.; Romero-Ocaña, I.; Fornasiero, P. H2 production by photocatalytic reforming of oxygenated compounds using TiO2-based materials. Mater. Sci. Semicond. Process. 2016, 42, 122 130.
(286) Li, X.; Cui, S.; Wang, D.; Zhou, Y.; Zhou, H.; Hu, Y.; Liu, J.; Long, Y.; Wu, W.; Hua, J.; Tian, H. New Organic Donor Acceptor Acceptor Sensitizers for Efficient Dye-Sensitized Solar Cells and Photocatalytic Hydrogen Evolution under Visible-Light Irradiation. ChemSusChem 2014, 7, 2879 2888.
(287) Dessi, A.; Calamante, M.; Mordini, A.; Peruzzini, M.; Sinicropi, A.; Basosi, R.; de Biani, F.; Taddei, M.; Colonna, D.; Di Carlo, A.; Reginato, G.; Zani, L. Organic dyes with intense light absorption especially suitable for application in thin-layer dye-sensitized solar cells. Chem. Commun. 2014, 50, 1395213955.
(288) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid metals. Phys. Rev. B 1993, 47, 558.
(289) Kresse, G.; Furthmüller, J. Efficiency of ab-initio total energy calculations for metals and
semiconductors using a plane-wave basis set. Comput. Mater. Sci. 1996, 6, 15 50.
(290) Kresse, G.; Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 1996, 54, 11169.
(291) Monkhorst, H. J.; Pack, J. D. Special points for Brillouin-zone integrations. Phys. Rev. B 1976, 13, 5188.
(292) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 1996, 77, 3865.
(293) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and accurate ab initio parametrization of
density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010, 132, 154104.
(294) Henkelman, G.; Jonsson, H. Improved tangent estimate in the nudged elastic band method for finding minimum energy paths and saddle points. J. Chem. Phys. 2000, 113, 9978 9985.
(295) Henkelman, G.; Uberuaga, B. P.; Jónsson, H. A climbing image nudged elastic band method for finding saddle points and minimum energy paths. J. Chem. Phys. 2000, 113, 9901 9904.
(296) Sheppard, D.; Terrell, R.; Henkelman, G. Optimization methods for finding minimum energy paths. J. Chem. Phys. 2008, 128, 134106.
(297) Sheppard, D.; Henkelman, G. Paths to which the nudged elastic band converges. J. Comput. Chem. 2011, 32, 1769 1771.